1 INFO-VAX	Thu, 06 Mar 2003	Volume 2003 : Issue 128       Contents: Re: %UAF-E-PWDSYNTAX Re: %UAF-E-PWDSYNTAX= Re: ?= Install patch vms731_sys_v0300 > v0200 and ACRTL_v010.  Re: Alpha doesn't boot Re: Alpha doesn't boot Re: Alpha doesn't boot6 Re: Anyone use a Compaq 20/40 dlt tape drive in a vax?P Re: Arguing about editors (was Re: EDT and command tables have invalid format (w AS2100 parts available Re: Carly's feedback link  Re: Carly's feedback link  Re: Carly's feedback link  Re: CLI question for HP reps/ Re: DIFFERENT BLOCKSIZE FOR BACKUP AND RESTORE? / Re: DIFFERENT BLOCKSIZE FOR BACKUP AND RESTORE? / Re: DIFFERENT BLOCKSIZE FOR BACKUP AND RESTORE? / Re: DIFFERENT BLOCKSIZE FOR BACKUP AND RESTORE? / Re: DIFFERENT BLOCKSIZE FOR BACKUP AND RESTORE? / Re: DIFFERENT BLOCKSIZE FOR BACKUP AND RESTORE? 6 Re: Editor wars - Was: Re: Upcoming VMS improvements ?6 Re: Editor wars - Was: Re: Upcoming VMS improvements ?6 Re: Editor wars - Was: Re: Upcoming VMS improvements ? Re: Emacs (again) on V7.3-1  RE: Emacs (again) on V7.3-1  Re: Fibre Channel and SCS  Re: Fibre Channel and SCS  Re: Fibre Channel and SCS  RE: Fibre Channel and SCS  Re: Fibre Channel and SCS  Re: Fibre Channel and SCS  Re: Fibre Channel and SCS  Re: Fibre Channel and SCS  Re: Fibre Channel and SCS  Re: Fibre Channel and SCS ( Free Support, All OSes, All Applications HP broke my link to WIS  Re: HP broke my link to WIS  Re: HP broke my link to WIS  Re: HP broke my link to WIS 4 Re: HP killing Golden Eggs- future of goose unknown.4 Re: HP killing Golden Eggs- future of goose unknown. iconv and UTF-8 on OpenVMS7.3-1 / Re: Linus trashes Itanium - he's scared of VMS! / Re: Linus trashes Itanium - he's scared of VMS! F Re: Maximum Record Size Error, -RMS-F-MRS, invalid maximum record sizeF Re: Maximum Record Size Error, -RMS-F-MRS, invalid maximum record size Re: Move Over, Alpha Down  Re: MS Virus< Open Source Benchmark Utility for Oracle and other Databases Re: oracle benchmarks on VMS Re: oracle benchmarks on VMS RE: oracle benchmarks on VMS. Re: OT Re: National Moratorium to Stop the War, Re: OT:  National Moratorium to Stop the War portability was/is: unix Re: Questions on SSH for VMS Re: Recent posts I have made' restoring a file from a backup (newbie) + Re: restoring a file from a backup (newbie) + Re: restoring a file from a backup (newbie) + Re: restoring a file from a backup (newbie)  RMS version number Re: RMS version number Re: RMS version number Re: RMS version number Re: RMS version number Re: RMS version number Re: RMS version number Re: RMS version number Re: RMS version number RE: RMS version number Re: RMS version number Re: RMS version number RE: RMS version number3 Re: SCSI cluster disk thrashing between connections 3 Re: SCSI cluster disk thrashing between connections  Re: unix Re: unix Re: unix Re: unix Re: unix Re: unix Re: unix Re: unix Re: unix/ Re: Upgrade to OPEN VMS 7.3-1 - Patches needed?  VAX again: unix 1 Re: VAXstation 4000 VLC VMS 5.5-2 Boot Parameters 1 Re: VAXstation 4000 VLC VMS 5.5-2 Boot Parameters 1 Re: VAXstation 4000 VLC VMS 5.5-2 Boot Parameters  Re: VMS Backup solutions Re: VMS Backup solutions Re: VMS Backup solutions Re: VMS Question???  Re: VMS Question???  Re: VMSTAR question  Re: [OT]:to Sue   F ----------------------------------------------------------------------  $ Date: Thu, 6 Mar 2003 10:02:15 +0100, From: "Rainer Giese" <waste.not@welcome.net> Subject: Re: %UAF-E-PWDSYNTAX 6 Message-ID: <b472rf$1t412j$1@ID-138444.news.dfncis.de>  L "Brian Tillman" <Brian.Tillman@notnoone.notnohow.com> schrieb im Newsbeitrag news:3e664a9c@news.si.com...  4 > Because AUTHORIZE doesn't pass errors back to DCL.   That's not generally true :    $ MCR AUTHORIZE MODIFY QWERT, %UAF-W-BADSPC, no user matches specification $ SHOW SYMBOL $STATUS    $STATUS == "%X13F18C80" % $ WRITE SYS$OUTPUT F$MESSAGE($STATUS) , %UAF-W-BADSPC, no user matches specification   Regards 
 Rainer  Giese    ------------------------------   Date: 6 Mar 2003 05:26:32 -0600 - From: Kilgallen@SpamCop.net (Larry Kilgallen)  Subject: Re: %UAF-E-PWDSYNTAX 3 Message-ID: <q+7aUiNlqi5H@eisner.encompasserve.org>   e In article <b472rf$1t412j$1@ID-138444.news.dfncis.de>, "Rainer Giese" <waste.not@welcome.net> writes:  > N > "Brian Tillman" <Brian.Tillman@notnoone.notnohow.com> schrieb im Newsbeitrag > news:3e664a9c@news.si.com... > 5 >> Because AUTHORIZE doesn't pass errors back to DCL.  >  > That's not generally true :   + But it is true often enough to be annoying.   D My impression is they are gradually fixing the cases where it fails,4 so reporting them through official channels is good.   ------------------------------  % Date: Thu, 06 Mar 2003 12:37:33 +0100 . From: "Hans M. Aus" <aus@vim.uni-wuerzburg.de>F Subject: Re: ?= Install patch vms731_sys_v0300 > v0200 and ACRTL_v010.B Message-ID: <aus-8DB56C.12373306032003@wrzx08.rz.uni-wuerzburg.de>   Thanks - worked just fine.   --  B Cheers, Hans M. Aus, Wuerzburg, Germany,  aus@vim.uni-wuerzburg.de   ------------------------------  % Date: Thu, 06 Mar 2003 08:40:12 +0000 ( From: Nic Clews <sendspamhere@127.0.0.1> Subject: Re: Alpha doesn't boot ) Message-ID: <3E67096C.DB94242B@127.0.0.1>    Horst Drechsel wrote:  > 	 > Hi all,  > @ >    one of our VMS (7.2-1) cluster members - AlphaPC164 500 MHz< > (chip 21164A) doesn't boot anymore. It comes to hang right/ > at the beginning of boot (over EWA0 adapter):  >  >    Try MOP boot  >    ...........? >    MOP booting sequence starts normally, server receives boot  >     request, says ...  > = >    %%%%%%%%%%%  OPCOM   5-MAR-2003 12:58:26.91  %%%%%%%%%%% % >    Message from user DECNET on A400 - >    DECnet event 0.3, automatic line service 4 >    From node 1.400 (A400),  5-MAR-2003 12:58:22.869 >    Circuit EWA-0, Load, Successful, Node = 1.16 (LUCKY) V >    File = A400$DKA0:<SYS15.>, Operating system, Ethernet address = 08-00-2B-E6-9B-5C >  >    client says ... >  >    (normal messages up to:)  >    ...( >    Setting affinity to the primary CPU >    Jumping to bootstrap mode > 2 >         >>>>>>>>>>  and here it hangs  <<<<<<<<<  F On your bootnode, run ANAL/DISK against the system disk and check that the backlinks are OK.   D Is service enabled on the bootnode for the ethernet adapter? I'm notH sure that the visible MOP request indicates that "service" is enabled orG not for the circuit. Check the parameters in NCP is you're using DECnet G or LANCP if using the new and better network boot (CLUSTER_CONFIG_LAN).   G It is possible there are more hardware errors. Have you tried booting a  distribution CDROM?    --  ? Regards, Nic Clews a.k.a. Mr. CP Charges, CSC Computer Sciences  nclews at csc dot com    ------------------------------  % Date: Thu, 06 Mar 2003 08:45:53 -0500 2 From: rdeininger@mindspring.com (Robert Deininger) Subject: Re: Alpha doesn't boot L Message-ID: <rdeininger-0603030845530001@user-uinj0ah.dialup.mindspring.com>  B In article <00A1C6AB.7B7C4C66.9@sternwarte.uni-erlangen.de>, Horst1 Drechsel <ai05@sternwarte.uni-erlangen.de> wrote:    >Hi all, > ? >   one of our VMS (7.2-1) cluster members - AlphaPC164 500 MHz ; >(chip 21164A) doesn't boot anymore. It comes to hang right . >at the beginning of boot (over EWA0 adapter): >  >   Try MOP boot >   ........... > >   MOP booting sequence starts normally, server receives boot >    request, says ...   > < >   %%%%%%%%%%%  OPCOM   5-MAR-2003 12:58:26.91  %%%%%%%%%%%$ >   Message from user DECNET on A400, >   DECnet event 0.3, automatic line service3 >   From node 1.400 (A400),  5-MAR-2003 12:58:22.86 8 >   Circuit EWA-0, Load, Successful, Node = 1.16 (LUCKY)C >   File = A400$DKA0:<SYS15.>, Operating system, Ethernet address =  08-00-2B-E6-9B-5C  >  >   client says ...  >  >   (normal messages up to:) >   ... ' >   Setting affinity to the primary CPU  >   Jumping to bootstrap mode  > 1 >        >>>>>>>>>>  and here it hangs  <<<<<<<<<    Does it really hang?  H How is the CONSOLE environment variable set?  These systems support bothH graphics and serial consoles, and I think this is one of the models thatI will respond to a keypress on the _inactive_ console connection.  You can G talk to the console, but when it hands off to the OS, output is only to F the "official" console port as specified by the environment variable. H IIRC, this happens at exactly the point where you are seeing the "hang".  I Fiddling with power supplies and other hardware often has the side effect # of resetting environment variables.    ------------------------------  + Date: Thu, 06 Mar 2003 14:49:51 +0100 (MET) 9 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com>  Subject: Re: Alpha doesn't boot ; Message-ID: <01KT7HWD87Z29H18VC@sysdev.deutsche-boerse.com>   3 > >        >>>>>>>>>>  and here it hangs  <<<<<<<<<  >  > Does it really hang? > J > How is the CONSOLE environment variable set?  These systems support bothJ > graphics and serial consoles, and I think this is one of the models thatK > will respond to a keypress on the _inactive_ console connection.  You can I > talk to the console, but when it hands off to the OS, output is only to H > the "official" console port as specified by the environment variable. J > IIRC, this happens at exactly the point where you are seeing the "hang". > K > Fiddling with power supplies and other hardware often has the side effect % > of resetting environment variables.   E Yes.  These used to really confuse me.  There is a console variable,  D graphics or serial, for the "real" console.  However, you can enter D commands on the other one.  Actually, I think that (in the case the G "real" console is the graphics one) you can get this alternate console  I on EITHER TT port, not just the one which can be OPA0 (usually or always   TTA0).   ------------------------------  % Date: Thu, 06 Mar 2003 09:37:59 -0500 ! From: Jim Agnew <jpagnew@vcu.edu> ? Subject: Re: Anyone use a Compaq 20/40 dlt tape drive in a vax? ' Message-ID: <3E675D46.2CB6B9D7@vcu.edu>   H Interesting... I have an Exabyte 8200 that came out of a device that wasA hung off a hsc controller.  when i put it on my vs3100m78, with a B storageworks box with 6 drives in it, it crashes vms..  not reallyG having time to fiddle, i noted it and moved on.. (im using it currently ? as a genderbender for two male scsi cables at the moment. ;-) )    Jim    Brian Tillman wrote: >  > Along the same lines...  > I > I have a DLT4000 currently attached to an HSD30.  On the VAX, the drive F > shows up as a TZ01.  It seems to work, although I realize it's not aI > supported configuration, at least the firmware docs for the HSD seem to  > indicate that. > M > I have another drive, a DLT2000 (effectively a TZ87), that I'd like to use. G > I daisy-chained it from the DLT4000 and it, too, showed up as a TZ01. M > However, neither would work reliably.  I'd get fatal controller errors and, J > once the DLT2000 hung and trying to reset the HSD caused a system crash.J > I'm pretty sure the DLT2000 is a working device, since it was on another  > platform and working correctly > J > I have attached to the SCSI bus on the VAX (KZQSA), a StorageWorks shelfM > containing two 8mm drives.  Could I daisy-chain the DLT2000 from the shelf? L > I'm thinking that if the device is directly attached to the VAX instead ofI > via the HSD, it might work more reliably.  I'm running OpenVMS VAX V7.2  > --K > Brian Tillman         Internet: Brian.Tillman at smiths-aerospace dot com 7 > Smiths Aerospace  Addresses modified to prevent SPAM. B > 3290 Patterson Ave. SE, MS Replace "at" with "@", "dot" with "." > Grand Rapids, MI 49512-1991 : >        This opinion doesn't represent that of my company   --  F "4,000 years ago I made a mistake."  Elrond Half-Elven, in "Fellowship of the Ring"   ------------------------------  % Date: Thu, 06 Mar 2003 19:52:12 +1000 1 From: Paddy O'Brien <paddy.o'brien@tg.nsw.gov.au> Y Subject: Re: Arguing about editors (was Re: EDT and command tables have invalid format (w , Message-ID: <3E671A4C.6030207@tg.nsw.gov.au>   Alan E. Feldman wrote:g > Paddy O'Brien <paddy.o'brien@tg.nsw.gov.au> wrote in message news:<3E646E18.2070607@tg.nsw.gov.au>...  >  >>Hoff Hoffman wrote: 	 >>[snips]  >> >>H >>>  I'm not particularly familiar with that EDT command syntax, and EDTH >>>  has not seen particular work in eons.  You've defined the DO key toI >>>  some sort of SPAWN, obviously.  (I can't replicate the SPAWN error.)  >>>  >>: >>  Whoa, don't get into an argument with Alan Feldman :-) >> >> >>  Regards, Paddy > G As you hopefully (the smilie) realised, I was just being facetious and   sortta jocular :-)   >  > G > Well, I get very frustrated when people misunderstand my posts. Maybe 0 > I wasn't being clear enough (I thought I was). > G Not intending any misunderstanding.  I use one editor on VMS because I  H was brought up with it.  I've used some pretty primitive ones from back I in the '60s.  In fact, my colleagues think my use of an editor is pretty  E primitive.  I use few "features" and just navigate my Fortran source  D codes in a very basic way.  I'm a happy dinsoaur like that :-)  Any H editor, to me, is just a simple tool to do my real work, and that's how H I like them -- simple.  In my editting, I happen not to have had a need E for such as your NOTRUNCATE, or some of the other "esoteric" (to me)  7 things in any editor mentioned by various contributors.   F > I get especially "annoyed" when I make it clear FROM THE OUTSET thatG > SHIFT LEFT and SHIFT RIGHT are not the same as having SET NOTRUNCATE. F > They are not even a good workaround for it. Yet, at least one posterE > (maybe two or three, I'd have to check) said ~"Well, just use SHIFT E > LEFT and SHIFT RIGHT!~" (~" = approx. quote) as if I had never said " > they were not a good substitute. > F >   AUUUUUUUUUUUUGGGGGGGGGGGGGGGGGGHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH! :-) > E > And then there's the "Well, you can write a TPU program!" response. G > Yeah, well I've already fixed what little I don't like about EDT, and H > I can do 95%-plus of my daily editing faster with EDT, so why should IG > bother? I am not fluent in TPU so there'd have to be a good reason to H > bother with it. I am glad it's there for times when I need it, but for! > all other editing I prefer EDT.  > ' > Is any of this so hard to understand?  > E > I do, however, greatly appreciate people's code to help me make EVE   > more efficient. Thank you all! > G > BTW, why did you quote Hoff's response which has almost nothing to do  > with any of this?  > F I only picked up his comment intended (as earlier) as a facetious and E jocular remark once he said that he was not familiar with EDT.  Sole  @ motivation and not any contribution to any other thread or as a . contribution for/against any of your comments.  C Bearing in mind that to me the functionality within EDT and TPU is  I largely not something I use, one thing I do like about TPU over EDT (and  I TPU happened to be my first VMS editor back in about '85) is the ability  F to use wildcards on the command line -- not an internal editor thing, H though.  And I think it came in later than when I first used the editor.   > [...]  >  > Disclaimer: JMNSHO > Alan E. Feldman      Regards, Paddy      G ***********************************************************************   C "This electronic message and any attachments may contain privileged > and confidential information intended only for the use of the B addressees named above.  If you are not the intended recipient of C this email, please delete the message and any attachment and advise B the sender.  You are hereby notified that any use, dissemination, 7 distribution, reproduction of this email is prohibited.   A If you have received the email in error, please notify TransGrid  A immediately.  Any views expressed in this email are those of the  = individual sender except where the sender expressly and with  C authority states them to be the views of TransGrid.  TransGrid uses > virus scanning software but excludes any liability for viruses contained in any attachment.  < Please note the email address for TransGrid personnel is now$ firstname.lastname@transgrid.com.au"  G ***********************************************************************    ------------------------------   Date: 6 Mar 2003 16:21:17 GMT ( From: bill@cs.uofs.edu (Bill Gunshannon) Subject: AS2100 parts available 6 Message-ID: <b47shs$1tgha0$1@ID-135708.news.dfncis.de>  . This is particularly for those who's motto is:A   They'll take my Alpha when they pry it from my cold dead hands.   C I have abbandoned any hope of repairing the dead AS2100 I had here. C I also don't have the space to continue to store the carcass, so, I D have striped it out pretty much.  Parts are available and no, not onC Ebay.  The problem was "The I/O Board", whatever that means.  So, I F have a (to the best of my knowledge) good power supply, the big motherB board (unless that's what they mean by I/O Board?), the little fanD controller board.  And, although I didn't pull them, if someone getsD back to me quickly (well, you probably have til tomorrow as the snowD is to deep to push this stuff to the dumpster :-) I would be willingC to pull the fans themselves out.  And, taking another quick look, I > could probably pull out the disk bay as well if it was needed.  G All of this can be had for the cost of shipping or if your close enough E (and can actually point to Scranton on a map) you are free to stop by 2 to pick up any or all of it.  I'll buy the coffee.  
 All the best.    bill   --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>       ------------------------------  # Date: Thu, 06 Mar 2003 14:02:46 GMT # From: "John Smith" <a@nonymous.com> " Subject: Re: Carly's feedback linkH Message-ID: <ayI9a.58028$em1.37090@news04.bloor.is.net.cable.rogers.com>  < "David J. Dachtera" <djesys.nospam@fsi.net> wrote in message! news:3E66BB17.D153BCA7@fsi.net... $ > The text of a message I just sent: > E > There is an extremely great, extremely urgent need to advertise VMS E > until it becomes as much second nature as "Winston tastes good like  a B > ..." I'm sure you can finish that even though cigarette ads have beenD > taboo on TV and radio for decades. VMS needs to be equally as well > known. > F > In this day and age of terrorism, hackers and crackers, the security and > > reliability of VMS may be our only defense against wholesale	 corporate  > sabotage.  > D > I have at least a dozen TV ads scripted and ready to shoot. I also haveE > print media ads that can be made camera ready on very short notice.  > F > VMS was once second only to IBM's MVS in the corporate world. Today, itA > is all but a memory. VMS is tremendously profitable as you well  know. B > Now is the time to tap that profit-center and raise VMS from the ashes  > of its former life.  > B > With the death of Alpha prior to the merger, that won't be easy.B > Promises like "bet your business on Alpha", have been repeatedly broken. C > Just mentioning "Compaq" in a sales presentation is enough to get  one B > thrown out of prospective client site. Just thinking "Alpha" can > endanger the sale. > ; > I can help. Please e-mail me to arrange a meeting at your  convenience.      A And the following is the only response you'll get back, as I have / found out by using the link to carly's mailbag:    "Dear xxxxxxxx:   E Thank you for taking the time to send HP your comments on mm/dd. They D have been forwarded to the appropriate people within Hewlett-Packard for E their information or action. Should more clarification or information  be& needed, you may be contacted directly.  8 Your input is important to us and very much appreciated.  
 Sincerely,   HP Executive Communications"    A By appropriate people, I'm sure they mean the guy who carries the  paper to the shredder.   ------------------------------  % Date: Thu, 06 Mar 2003 08:05:06 -0600  From: brandon@dalsemi.com " Subject: Re: Carly's feedback link1 Message-ID: <03030608050673@dscis6-0.dalsemi.com>   C > And the following is the only response you'll get back, as I have 1 > found out by using the link to carly's mailbag:  >  > "Dear xxxxxxxx:  > G > Thank you for taking the time to send HP your comments on mm/dd. They F > have been forwarded to the appropriate people within Hewlett-Packard > for G > their information or action. Should more clarification or information  > be( > needed, you may be contacted directly. > : > Your input is important to us and very much appreciated. >  > Sincerely, >  > HP Executive Communications" >  > C > By appropriate people, I'm sure they mean the guy who carries the  > paper to the shredder. >   M Reads like the e-mail I would get when I was on an Internet dating service... 9 The "Dear John" letters - wait a tick, my name is John...    John Brandon VMS Systems Administrator  Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wk  972.371.4003 fx    ------------------------------  % Date: Thu, 06 Mar 2003 09:43:34 -0500 + From: Steve Lionel <Steve.Lionel@intel.com> " Subject: Re: Carly's feedback link8 Message-ID: <5fne6vcgjvitbu2th1en732olleqdnfvue@4ax.com>  F On Thu, 06 Mar 2003 14:02:46 GMT, "John Smith" <a@nonymous.com> wrote:  B >And the following is the only response you'll get back, as I have0 >found out by using the link to carly's mailbag:   snip  B >By appropriate people, I'm sure they mean the guy who carries the >paper to the shredder.   D Don't be so sure.  A couple of weeks ago, I got contacted by Carly'sD office to help resolve a customer problem with Compaq Visual FortranF that had somehow gotten there - though the customer had never botheredC to ask CVF support.  And the problem was that the customer had lost  his product CD...   A While you may get a form letter in response initially, there IS a E staff that looks into each requests and tries to do the right thing.     Steve   D Please send Visual Fortran support requests to vf-support@compaq.com   Steve Lionel Software Products Division Intel Corporation 
 Nashua, NH  = User communities for Intel Fortran and Compaq Visual Fortran:     http://intel.com/IDS/community   ------------------------------   Date: 6 Mar 2003 11:20:55 -0600 ; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) % Subject: Re: CLI question for HP reps 3 Message-ID: <hqSuqq+Mi6CS@eisner.encompasserve.org>   k In article <VCw9a.528$vM1.101887@news.uswest.net>, "Michael D. Ober" <obermd-@-alum-mit-edu-nospam> writes: I > You only need to do this once for each utility you want to convert.  If ; > you're too lazy to do this, don't use the unix utilities.   H    I admit I'm lazy, but I wasn't the one asking for more eunichs stuff.6    I wish I never had to use any of this eunichs crap!   ------------------------------   Date: 6 Mar 2003 00:06:35 -0800 % From: Bart.Zorn@xs4all.nl (Bart Zorn) 8 Subject: Re: DIFFERENT BLOCKSIZE FOR BACKUP AND RESTORE?= Message-ID: <a98cd882.0303060006.3785998c@posting.google.com>   F The blocksize qualifier is ignored for restores. In stead, BACKUP uses6 the information that is present in the saveset header.  E You may however expect problems, sometimes. Using /NOCRC and /GROUP=0 < is a very unwise thing to do. Neither your tapedrive nor theB cartridges that you use are perfect. Those two error detecting andC error recovery options may come in handy if you ever need them. The C penalty of using more CPU time during the backup is certainly worth A it. You will know once you try to restore a tape without them and C there are errors. Remember that a tape that is perfectly restorable ! now, may not be in a year's time.   : On a related issue: I suppose you also do not use /VERIFY?  	 Bart Zorn     i Kiasu Surfer <verykiasu@hotpop.com> wrote in message news:<8rgd6v0e4afpegpahubqm73fhjvhqrknhr@4ax.com>... Q > I wonder if the different blocksize used for backup and restore would cause any ' > data integrity problem after restore?  > P > I am considering using BLOCKSIZE of 61440 for my DLT IV tape as recommended byG > Compaq FAQ web, but presently all my backup using BLOCKSIZE of 32256.  > R > If i were to change restore scripts, would restoration of the older backup tapesN > be any problem? (OVMS 7.1 on Alpha, backup using /NOCRC /GROUP=0 parameters) >  > Thanks in advance.
 > Regards, >  > Kevin Lai  >  > **************** > ** SPAM BLOCK ** > ****************A > REPLACE "verykiasu" with "iamverykiasu" at verykiasu@hotpop.com    ------------------------------  % Date: Thu, 06 Mar 2003 08:29:30 +0000 ( From: Nic Clews <sendspamhere@127.0.0.1>8 Subject: Re: DIFFERENT BLOCKSIZE FOR BACKUP AND RESTORE?) Message-ID: <3E6706EA.D6811C0D@127.0.0.1>    Kiasu Surfer wrote:  > Q > I wonder if the different blocksize used for backup and restore would cause any ' > data integrity problem after restore?  > P > I am considering using BLOCKSIZE of 61440 for my DLT IV tape as recommended byG > Compaq FAQ web, but presently all my backup using BLOCKSIZE of 32256.  > R > If i were to change restore scripts, would restoration of the older backup tapesN > be any problem? (OVMS 7.1 on Alpha, backup using /NOCRC /GROUP=0 parameters)  E Bart is right to point out, particularly when you use the words "data H integrity" while quoting a command that has "GROUP=0" and "NOCRC" in it,@ they are mutually exclusive. Having said that, modern DLT drivesD typically do have hardware CRC, but without a verification pass, you@ could quite easily have written something totally unrecoverable.  G One factor is that you would be unable to copy a saveset from tape onto E disk, maximum size on disk is 32768 (ish) and to move savesets around F you'll need Save Set Manager (layered product), assuming your tape was	 readable.   F Sorry to sound so negative, but when you've been around long enough toG find your cynicism in in computer hardware proved, you tend to approach E jobs with care and caution, particularly when it isn't mine [data] to  lose.    --  ? Regards, Nic Clews a.k.a. Mr. CP Charges, CSC Computer Sciences  nclews at csc dot com    ------------------------------   Date: 6 Mar 2003 05:22:16 -0600 - From: Kilgallen@SpamCop.net (Larry Kilgallen) 8 Subject: Re: DIFFERENT BLOCKSIZE FOR BACKUP AND RESTORE?3 Message-ID: <pSD$pfUxdIh1@eisner.encompasserve.org>   e In article <a98cd882.0303060006.3785998c@posting.google.com>, Bart.Zorn@xs4all.nl (Bart Zorn) writes: H > The blocksize qualifier is ignored for restores. In stead, BACKUP uses8 > the information that is present in the saveset header. > G > You may however expect problems, sometimes. Using /NOCRC and /GROUP=0 > > is a very unwise thing to do. Neither your tapedrive nor theD > cartridges that you use are perfect. Those two error detecting andE > error recovery options may come in handy if you ever need them. The E > penalty of using more CPU time during the backup is certainly worth C > it. You will know once you try to restore a tape without them and E > there are errors. Remember that a tape that is perfectly restorable # > now, may not be in a year's time.  > < > On a related issue: I suppose you also do not use /VERIFY?  K For anyone who wants to save X% of their resources with /NOCRC and /GROUP=0 J ans also wants to save Y% of their resources by skipping /VERIFY, I have a+ way to save 100% of their backup resources.   + After all, making backups is not important.   9 It is only the ability to _restore_ backups that matters.    ------------------------------   Date: 6 Mar 2003 09:18:29 -0600  From: briggs@encompasserve.org8 Subject: Re: DIFFERENT BLOCKSIZE FOR BACKUP AND RESTORE?3 Message-ID: <g+I1Q6fA9GW5@eisner.encompasserve.org>   c In article <pSD$pfUxdIh1@eisner.encompasserve.org>, Kilgallen@SpamCop.net (Larry Kilgallen) writes: M > For anyone who wants to save X% of their resources with /NOCRC and /GROUP=0 L > ans also wants to save Y% of their resources by skipping /VERIFY, I have a- > way to save 100% of their backup resources.  > - > After all, making backups is not important.  > ; > It is only the ability to _restore_ backups that matters.   J This tends to be a bit of a religious issue.  Larry certainly has a point.  A On the other hand, /CRC doesn't check all that much more than the B drive firmware already does.  But if you have the CPU to spare, it, would be a foolish economy to turn off /CRC.  > /GROUP=0 is almost exactly as safe as /GROUP=10 on many modernG drives (if you get an error on a block, chances are extremely good that * you won't be able to read past it anyway).  G /VERIFY...  Well, it's hard to argue against using /VERIFY.  But you're C taking at least a 2 to 1 performance hit with /VERIFY.  And in some @ cases the penalty can even go beyond 10 to 1 (with multiple saveA sets on a single tape, repositioning to the beginning of the save  set can get nasty).   D Whether or not you use /VERIFY, you still need to attempt at least aF dry run of the restore procedures every now and again.  That can catch things that even /VERIFY won't.    	John Briggs   ------------------------------   Date: 6 Mar 2003 07:36:22 -0700 1 From: nothome@spammers.are.scum (Malcolm Dunnett) 8 Subject: Re: DIFFERENT BLOCKSIZE FOR BACKUP AND RESTORE?- Message-ID: <gWcgnya0sStg@malvm7.mala.bc.ca.>   4 In article <g+I1Q6fA9GW5@eisner.encompasserve.org>, $     briggs@encompasserve.org writes:   > C > On the other hand, /CRC doesn't check all that much more than the D > drive firmware already does.  But if you have the CPU to spare, it. > would be a foolish economy to turn off /CRC. > A   /CRC is a better end-to-end check. I've seen problems in memory @ controllers and PCI busses that can cause silent data corruptionA while the data is travelling from memory to the tape drive. a CRC ? in the tape drive won't catch this because by the time the tape A drive calculates its CRC the data is already corrupted. Of course > these sort of problems are extremely rare ( I've seen it twice> in 25 years ) but the cost of calculating CRC on a fast CPU is pretty low.   I > /VERIFY...  Well, it's hard to argue against using /VERIFY.  But you're E > taking at least a 2 to 1 performance hit with /VERIFY.  And in some B > cases the penalty can even go beyond 10 to 1 (with multiple saveC > sets on a single tape, repositioning to the beginning of the save  > set can get nasty).  >   ?   With multiple savesets on a tape it can be quite a bit faster > to write them all first then do a /COMPARE pass against all of= them afterwards ( in the same order they were written ). This < prevents having to seek to the beginning of each saveset for= the /VERIFY. The downside is that you may get bogus VERIFYERR ; errors because the data on disk may have changed after your > backup finished and before you do the COMPARE ( of course this= can happen with /VERIFY too, but the window is smaller if you  verify right away ).   ------------------------------   Date: 6 Mar 2003 09:55:54 -0800 . From: spamsink2001@yahoo.com (Alan E. Feldman)8 Subject: Re: DIFFERENT BLOCKSIZE FOR BACKUP AND RESTORE?= Message-ID: <b096a4ee.0303060955.48908a10@posting.google.com>   Y Nic Clews <sendspamhere@127.0.0.1> wrote in message news:<3E6706EA.D6811C0D@127.0.0.1>...  > Kiasu Surfer wrote:  > > S > > I wonder if the different blocksize used for backup and restore would cause any ) > > data integrity problem after restore?  > > R > > I am considering using BLOCKSIZE of 61440 for my DLT IV tape as recommended byI > > Compaq FAQ web, but presently all my backup using BLOCKSIZE of 32256.  > > T > > If i were to change restore scripts, would restoration of the older backup tapesP > > be any problem? (OVMS 7.1 on Alpha, backup using /NOCRC /GROUP=0 parameters) >  [...] I > One factor is that you would be unable to copy a saveset from tape onto G > disk, maximum size on disk is 32768 (ish) and to move savesets around H > you'll need Save Set Manager (layered product), assuming your tape was > readable.   D The maximum block size for a save set on disk is 32256 bytes. If youF have a save set on tape whose block size is larger than that, then you! cannot copy the save set to disk.    [...]    Disclaimer: JMHO Alan E. Feldman    ------------------------------  % Date: Thu, 06 Mar 2003 09:38:59 -0500 ! From: Jim Agnew <jpagnew@vcu.edu> ? Subject: Re: Editor wars - Was: Re: Upcoming VMS improvements ?	' Message-ID: <3E675D83.A1529DD1@vcu.edu>   H were thos the hazeltines with a big cpu box under it about the size of aE pizza box vs3100, a little "neck", and then the crt box above it, all' one piece, with CORE memory???   jimo  
 ualski wrote:0 >  > Paul Sture wrote:a >  > >r- > > Anyone remember SOS, whilst we are at it?b > > -- > > Paul Sture > K > I loved SOS after using TECO, not that I admited that in public. That and J > those Hazeltine terminals with neon lamps under the tall push-buttons on4 > the keyboard where a big improvement over KSR/ASR. >  > -- Aaron Sliwinski   --  F "4,000 years ago I made a mistake."  Elrond Half-Elven, in "Fellowship of the Ring"   ------------------------------  $ Date: Thu, 6 Mar 2003 10:25:28 -0500  From: John Santos <JOHN@egh.com>? Subject: Re: Editor wars - Was: Re: Upcoming VMS improvements ?t6 Message-ID: <1030306102300.26948A-100000@Ives.egh.com>  % On 5 Mar 2003, Alan E. Feldman wrote:n  X > Don Sykes <alphase@pacbell.net> wrote in message news:<3E6647FB.A21A11@pacbell.net>... > > "Alan E. Feldman" wrote: > > >  > [...] A > > > EVE fan: You can have multiple buffers in multiple windows!e > > > : > > > EDT fan: Well, in EDT you can have multiple buffers. > >  > > You can in TPU also. >  > * > Yes, I just said that. Let me try again: >  > % > EDT: You can have multiple buffers!  > G > EVE: You can have multiple buffers *and* see more than one of them ate > the same time! > # > EDT: Well, what about long lines?s > F > EVE: I can show you long lines. Just use SHIFT LEFT and SHIFT RIGHT! > H > EDT: I can show you long lines *and* you can see the whole line at the* > same time because I have SET NOTRUNCATE! >  >  -- OR --o > G > EVE: Well, with EDT you have to switch buffers to see another buffer.zC > EDT: Well, with EVE you have to SHIFT to see the next part of thea > line.e   EVA: Times Square!   ED: Fresh Air!   EVA: The Stores!   ED: The Chores!r   >  > [...] G > > > Hopefully this makes it clear for people who don't understand theT > > > value of SET NOTRUNCATE. > > K > > It is VERY CLEAR that the SET NOTRUNCATE is important to you. If (when)e > G > Well, I thought this comparison between multiple buffers and multipleo< > screen-width sections of a line would clarify it for some. > L > > I get the time I'll see if I can solve that one. (too busy now trying to? > > get SMTP up on my NT box - see post Move Over, Alpha Down).M > H > Please, there's no rush. If you can do it, great. Just post it or send: > me an e-mail (yes, the spamsink address is real) Thanks. >  > [...]o >  > Disclaimer: JMHO > Alan E. Feldmane >  >    -- e John Santos  Evans Griffiths & Hart, Inc. 781-861-0670 ext 539   ------------------------------  $ Date: Thu, 6 Mar 2003 11:23:14 -0500- From: "Peter Weaver" <peter.weaver@stelco.ca> ? Subject: Re: Editor wars - Was: Re: Upcoming VMS improvements ?d6 Message-ID: <b47slp$1roe8v$1@ID-141708.news.dfncis.de>   Alan E. Feldman wrote: >...F > With EDT it takes approx. 0.0 seconds. With EVE I get 2 to 3 secondsE > of unreadably fast messages in sequence at the bottom of the screenlA > with the cursor moving left and right across it and the message F > changing from normal to reverse video and back a few times. Why even >...  M Don't know what these messages are, I don't see them, is it possible that youh' customization added that is doing this?e  H > reaches the top, the cursor is still not at the top. The cursor has toG > wait for the message at the bottom of the screen to flash "Attempt tolB > move past the beginning of buffer AEFL.COM". First it appears in >...  D Not that I want you to give up EDT and move to EVE, but try this andD it may help you the next time you have to use EVE for something that EDT doesn't do as well.a   $! $ create sys$login:eve$edt.dif -  255,  258 !    if current_offset = 0	 !    thenTO !       move_vertical (-1);                     ! Output error message if @ bob  !    endif;- -  311,  314) !if mark (NONE) = end_of (current_buffer)m !thenl1 !    move_vertical (1);  ! force an error messageA !endif;N -  606,  609) !if mark (NONE) = end_of (current_buffer)  !thena0 !    move_vertical (1); ! force an error message !endif;G -  702,  7023 !       move_vertical (-1);     ! for error messagen - 1048, 1051) !if mark (NONE) = end_of (current_buffer)  !then6- !    move_vertical (1); ! force error messaged !endif;s /r $!% $ EDIT/SUM SYS$EXAMPLES:EVE$EDT.TPU -l2     /UPDATE=(SYS$LOGIN:EVE$EDT.DIF)/out=sys$login: $!= $ edit/tpu/nodisplay /command=sys$input sys$login:eve$edt.tpue eve_extend_tpu ("*");g+ eve_save_extended_tpu("sys$login:new_eve");- $!   P.S.; Did you ever try my attempt at the NOTRUNCATE that I postedr a few days ago?    -- Peter WeaverD Opinions are my own, and do not reflect the opinions of my employer,A nor the company that it sub-contracts to, nor the company that itz sub-contracts to.-   ------------------------------   Date: 6 Mar 2003 09:57:05 -0800 ( From: gtackett@yahoo.com (Galen Tackett)$ Subject: Re: Emacs (again) on V7.3-1= Message-ID: <1579dafc.0303060607.73012a64@posting.google.com>   ! > http://vms.gnu.org/history.htmls2 > http://vms.gnu.org/software/released1/emacs.html  D Unfortunately when I try the FTP sites mentioned there I am told (ifE via a browser) that the document contains no data; if I try using them/ FTP application, my connection never completes.r  B I think the problem has to do with the firewalls here, or possiblyF other security mechanisms (there are a lot here) that I'm unaware of.  Sigh...      Any other sites I might try?    X Z  <zarlenga@conan.ids.net> wrote in message news:<v66j5vcl6jk1f7@corp.supernews.com>...+ > Galen Tackett <gtackett@yahoo.com> wrote:4E > : Can anyone point me to a site where I can download Emacs V19.x (IlG > : believe that was the most recent Emacs ported to VMS) that will run C > : okay on VMS V7.3-1? The old DECUS sites are, I guess, no longer > > : working, and the "official" Gnu sites are also failing me. > ! > http://vms.gnu.org/history.htmlt2 > http://vms.gnu.org/software/released1/emacs.html   ------------------------------  $ Date: Thu, 6 Mar 2003 10:02:31 -0800# From: "Tom Linden" <tom@kednos.com> $ Subject: RE: Emacs (again) on V7.3-19 Message-ID: <CIEJLCMNHNNDLLOOGNJIKEAAGMAA.tom@kednos.com>    ftp.nvg.ntnu.no:/pub/vms/emacs/      >-----Original Message-----C0 >From: Galen Tackett [mailto:gtackett@yahoo.com]' >Sent: Thursday, March 06, 2003 9:57 AMt >To: Info-VAX@Mvb.Saic.Com% >Subject: Re: Emacs (again) on V7.3-1a >a >e" >> http://vms.gnu.org/history.html3 >> http://vms.gnu.org/software/released1/emacs.html  > E >Unfortunately when I try the FTP sites mentioned there I am told (if-F >via a browser) that the document contains no data; if I try using the0 >FTP application, my connection never completes. >RC >I think the problem has to do with the firewalls here, or possibly-G >other security mechanisms (there are a lot here) that I'm unaware of. r >Sigh... >e >e >Any other sites I might try?a >. >-. >Z  <zarlenga@conan.ids.net> wrote in message , >news:<v66j5vcl6jk1f7@corp.supernews.com>..., >> Galen Tackett <gtackett@yahoo.com> wrote:F >> : Can anyone point me to a site where I can download Emacs V19.x (IH >> : believe that was the most recent Emacs ported to VMS) that will runD >> : okay on VMS V7.3-1? The old DECUS sites are, I guess, no longer? >> : working, and the "official" Gnu sites are also failing me.  >> u" >> http://vms.gnu.org/history.html3 >> http://vms.gnu.org/software/released1/emacs.html0 >  >---' >Incoming mail is certified Virus Free.a; >Checked by AVG anti-virus system (http://www.grisoft.com).1A >Version: 6.0.449 / Virus Database: 251 - Release Date: 1/27/2003  >  ---t& Outgoing mail is certified Virus Free.: Checked by AVG anti-virus system (http://www.grisoft.com).@ Version: 6.0.449 / Virus Database: 251 - Release Date: 1/27/2003   ------------------------------  % Date: Thu, 06 Mar 2003 06:29:39 -0600t From: brandon@dalsemi.comm" Subject: Re: Fibre Channel and SCS1 Message-ID: <03030606293955@dscis6-0.dalsemi.com>t  / > We tried memory channel and that caused cpu 0o? > overloads due to lack of fastpath support for memory channel.o  F I have a small cluster (2 DS20 and 1 ES40) using MC-II; I am seriouslyM considering replacing with a private 100 Mbit network - for similiar reasons,oM well, maybe all conjecture on my part.  Anyone have objections, observations,e thoughts on this?c    F > The SCS traffic should never be on the network or disk subsystems if > possible,0E > just clutters it with messages used only by VMS clusters. It works,dG > but why not provide a dedicated SCS port for each node...and not deal7! > with a possible network outage.e  , You run SCS over a private network, correct?  F > We use the stop_bus procedure in sys$examples to turn off SCS on theE > network ports we do not use for cluster traffic. We do use FDDI and F > GIGe for the NI connections between the CI clusters, but one side isD > used only for shadow copies, as we currently cannot drive the high > lock traffic across the NI.o  G At a DFW-Days held here a few years ago, we had an engineer present hisdH comparitive studies of MC-II, 10/100, GIGe, DSSI, etc. of performance ofN various simulations.  100 and GIGe out performed MC-II in a number of facets. # Hence my drive to get off of MC-II.   K My simple plan is to replace MC with a 100 Mbit private network and use the , existing GIGe as a primary/failover for SCS.   John Brandon VMS Systems Administrator  Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wkn 972.371.4003 fxr   ------------------------------  % Date: Thu, 06 Mar 2003 08:16:04 -0500E2 From: rdeininger@mindspring.com (Robert Deininger)" Subject: Re: Fibre Channel and SCSL Message-ID: <rdeininger-0603030816040001@user-uinj0ah.dialup.mindspring.com>  L In article <03030606293955@dscis6-0.dalsemi.com>, brandon@dalsemi.com wrote:  0 >> We tried memory channel and that caused cpu 0@ >> overloads due to lack of fastpath support for memory channel. >hG >I have a small cluster (2 DS20 and 1 ES40) using MC-II; I am seriously N >considering replacing with a private 100 Mbit network - for similiar reasons,N >well, maybe all conjecture on my part.  Anyone have objections, observations, >thoughts on this?  J Unless you are out of PCI slots, you might as well keep the MC and add 100 Mb networking.  Use both.e    H >At a DFW-Days held here a few years ago, we had an engineer present hisI >comparitive studies of MC-II, 10/100, GIGe, DSSI, etc. of performance ofoO >various simulations.  100 and GIGe out performed MC-II in a number of facets.  $ >Hence my drive to get off of MC-II.  I You shouldn't make decisions based on those presentations.  Take the time4C to understand your workload.  Find your bottleneck(s).  Many of the 0 slideware tests use pretty artificial workloads.  B MC is still at or near the top of the list for low latency.  NewerD interconnects have more bandwidth, and the latest GbE stuff has goodJ enough latency to make the difference uninteresting in most cases.  GbE isG also likely cheaper, but that doesn't matter if you already have the MCi	 hardware.   L >My simple plan is to replace MC with a 100 Mbit private network and use the- >existing GIGe as a primary/failover for SCS.   I The new GbE adapters (DEGXA, if I haven't scrambled the name) ought to be H considered for any new purchases.  They are way better than the previousJ generation.  Even if you only have 100Mb network switches, the Gb adapters, might be worthwhile for later expandability.  F If you're running anything older than V7.3-1, a software upgrade couldJ well be your easiest path to better performance.  Again, it depends on the
 workloads.   ------------------------------  % Date: Thu, 06 Mar 2003 09:45:05 -0600, From: brandon@dalsemi.com " Subject: Re: Fibre Channel and SCS1 Message-ID: <03030609450586@dscis6-0.dalsemi.com>s  L > Unless you are out of PCI slots, you might as well keep the MC and add 100 > Mb networking.  Use both.n  M The DS20 is a bit lean on the PCI slots and I have to go through that part ofe it and make some decisions.w  J > >At a DFW-Days held here a few years ago, we had an engineer present hisK > >comparitive studies of MC-II, 10/100, GIGe, DSSI, etc. of performance ofaP > >various simulations.  100 and GIGe out performed MC-II in a number of facets.& > >Hence my drive to get off of MC-II. >.< > You shouldn't make decisions based on those presentations.  O Actually, I value the information I was given and do base decisions on educated N material.  And considering where the information comes from - that elevates my? level of attention to it.  I consider the soure to be reliable.n   >  Take the time8 > to understand your workload.  Find your bottleneck(s).  K The problem is not a bottleneck so much of MC, it is the performance of themM DS20 - well at least in the production environment it is in.  Any performancemL increase in the CPU arena will be appreciated.  Granted this is an irregularK configuration (in my mind), two DS20 and one ES40, but it was dropped in myk lap.    N This is not to make mention of the isntalled GIGe cards that are piped (at theF switch) to a 10-Mbit - that is correct, ten Mbit.  Other issue here...   >  Many of the2 > slideware tests use pretty artificial workloads.  L Artificial workloads?  Hmmm, every performance spec that I have seen for anyM CPU has been artificial.  And though your statement is true, and I agree thatTM there are implications associated with this, it is the only solid informationw> that we have to make our decissions on.  That and a few beers.  D > MC is still at or near the top of the list for low latency.  NewerF > interconnects have more bandwidth, and the latest GbE stuff has goodL > enough latency to make the difference uninteresting in most cases.  GbE isI > also likely cheaper, but that doesn't matter if you already have the MC! > hardware.    The data I have shows -   For a single stream workload test scenario: /     MC2 with a low-latency of 25% over  100Mbitt/     MC2 with a through-put of 20% over  100Mbit /     MC2 with a CPU cost    of 22% under 100Mbit   4   For a multiple stream (32) workload test scenario:/     100Mbit with a through-put of 25% over  MC2 &     no low-latency test data available/     100Mbit with a CPU cost    of 28% over  MC2e  O Even though this data is artificial, it does demonstrate the differences of MC2dF and 100Mbit.  The GIGe demonstrated similiar traits, but better in theH through-put arena.  And you are correct, the GIGe has better throughput.  N Based on the above information, in a heavy workload environment, 100Mbit would2 out-perform MC2 for both through-put and CPU cost.  M I remember when MC came out.  Touted at 100Mbyte throughput (actually only 60nK or 80 at the time) with a path to 200Mbyte (MC2).  At the time, at least of O what I remember, there was nothing that could compare to it.  I believe we were  still in the 10Mbit arena.  N > >My simple plan is to replace MC with a 100 Mbit private network and use the/ > >existing GIGe as a primary/failover for SCS.h >wK > The new GbE adapters (DEGXA, if I haven't scrambled the name) ought to be J > considered for any new purchases.  They are way better than the previousL > generation.  Even if you only have 100Mb network switches, the Gb adapters. > might be worthwhile for later expandability.  B I will not be able to justify a GIGe purchase against a 100Mbit.    M And considering my SCS traffic on my three node cluster (SAN's) a MC is a bitnK overkill and much more costly than what I can do with a private 100Mbit & asB GIGe connections, acting both as primary and failover SCS traffic.  H > If you're running anything older than V7.3-1, a software upgrade couldL > well be your easiest path to better performance.  Again, it depends on the > workloads.  ; We are On V7.2-1.  Working on the upgrade justification... e   Thanks for the feedback!         John Brandon VMS Systems Administrator  Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wk  972.371.4003 fx    ------------------------------  $ Date: Thu, 6 Mar 2003 11:36:42 -0500' From: "Main, Kerry" <Kerry.Main@hp.com>*" Subject: RE: Fibre Channel and SCST Message-ID: <BE56C50EA024184DAF48F0B9A47F5CF402660DF6@kaoexc01.americas.cpqcorp.net>   John,e  D As a suggestion, one option with MC not on systems might be to use 2D DE602's (dual port 100MbE) on each system - one on each separate PCIG bus. Dedicate 1 port from each DE602 to load balance SCS traffic across H PCI buses and connect each port to different Network switch. Use the newH SCACP utility in V7.3-1 to balance the traffic and set the costs on each path.   H If using TCPware or Multinet, you can set up virtual IP on the other twoG ports (one VIP, two physical ports) for Telnet and FTP traffic for highoG availability of Telnet and FTP traffic. This allows you to lose a cablesG (pull cable to simulate) and telnet and ftp sessions continue without a  hiccup.h  H On the Network switch, if it supports it, setup two separate VLANS - oneC for SCS cluster traffic and another for normal TCPIP / Ftp / DECnett traffic. =20  E We are doing this on a cluster of 3 ES45's using TCPware and it workseH great. Actually we have three VLANS - an additional one for Mgmt traffic as well.  D We are planning to take advantage of cluster scs traffic over the FCE HBA's in the future as a "backup" path that would only be used in the B event that the two network switches both failed. This gives us theF capability to have the cluster continue (batch, local access etc) even if both network switches died.   Regardsu  
 Kerry Main Senior Consultantt Hewlett-Packard (Canada) Co.! Consulting & Integration Servicesa Voice: 613-592-4660  Fax   : 613-591-4477 Email: kerryDOTmain@hpDOTcom-     (remove the DOT's and replace with "."'s)r OpenVMS DCL - the original .COM    -----Original Message-----9 From: brandon@dalsemi.com [mailto:brandon@dalsemi.com]=20n Sent: March 6, 2003 10:45 AM To: Info-VAX@Mvb.Saic.Comm" Subject: Re: Fibre Channel and SCS    H > Unless you are out of PCI slots, you might as well keep the MC and add   > 100 Mb networking.  Use both.g  E The DS20 is a bit lean on the PCI slots and I have to go through thatW# part of it and make some decisions.I  I > >At a DFW-Days held here a few years ago, we had an engineer present=20TC > >his comparitive studies of MC-II, 10/100, GIGe, DSSI, etc. of=20 H > >performance of various simulations.  100 and GIGe out performed MC-II  = > >in a number of facets. Hence my drive to get off of MC-II.a > < > You shouldn't make decisions based on those presentations.  F Actually, I value the information I was given and do base decisions onF educated material.  And considering where the information comes from -F that elevates my level of attention to it.  I consider the soure to be	 reliable.m   >  Take the time8 > to understand your workload.  Find your bottleneck(s).  G The problem is not a bottleneck so much of MC, it is the performance oftE the DS20 - well at least in the production environment it is in.  AnylH performance increase in the CPU arena will be appreciated.  Granted thisF is an irregular configuration (in my mind), two DS20 and one ES40, but it was dropped in my lap. =20s  F This is not to make mention of the isntalled GIGe cards that are piped (at theiF switch) to a 10-Mbit - that is correct, ten Mbit.  Other issue here...   >  Many of the2 > slideware tests use pretty artificial workloads.  H Artificial workloads?  Hmmm, every performance spec that I have seen forF any CPU has been artificial.  And though your statement is true, and IF agree that there are implications associated with this, it is the onlyE solid information that we have to make our decissions on.  That and aa
 few beers.  G > MC is still at or near the top of the list for low latency.  Newer=20 I > interconnects have more bandwidth, and the latest GbE stuff has good=20sH > enough latency to make the difference uninteresting in most cases. =20G > GbE is also likely cheaper, but that doesn't matter if you already=20r > have the MC hardware.c   The data I have showsi-   For a single stream workload test scenario:i/     MC2 with a low-latency of 25% over  100Mbite/     MC2 with a through-put of 20% over  100Mbitg/     MC2 with a CPU cost    of 22% under 100Mbitx  4   For a multiple stream (32) workload test scenario:/     100Mbit with a through-put of 25% over  MC2 &     no low-latency test data available/     100Mbit with a CPU cost    of 28% over  MC2S  H Even though this data is artificial, it does demonstrate the differencesF of MC2 and 100Mbit.  The GIGe demonstrated similiar traits, but betterC in the through-put arena.  And you are correct, the GIGe has better  throughput.e  H Based on the above information, in a heavy workload environment, 100Mbit8 would out-perform MC2 for both through-put and CPU cost.  E I remember when MC came out.  Touted at 100Mbyte throughput (actually G only 60 or 80 at the time) with a path to 200Mbyte (MC2).  At the time,cH at least of what I remember, there was nothing that could compare to it., I believe we were still in the 10Mbit arena.  I > >My simple plan is to replace MC with a 100 Mbit private network and=20j7 > >use the existing GIGe as a primary/failover for SCS.b >hH > The new GbE adapters (DEGXA, if I haven't scrambled the name) ought to  G > be considered for any new purchases.  They are way better than the=20lH > previous generation.  Even if you only have 100Mb network switches,=20> > the Gb adapters might be worthwhile for later expandability.  D I will not be able to justify a GIGe purchase against a 100Mbit. =20  G And considering my SCS traffic on my three node cluster (SAN's) a MC isbE a bit overkill and much more costly than what I can do with a privatetE 100Mbit & a GIGe connections, acting both as primary and failover SCSs traffic.  H > If you're running anything older than V7.3-1, a software upgrade could  H > well be your easiest path to better performance.  Again, it depends on   > the workloads.  = We are On V7.2-1.  Working on the upgrade justification...=20    Thanks for the feedback!         John Brandon VMS Systems Administratoro Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wkC 972.371.4003 fxr   ------------------------------   Date: 6 Mar 2003 09:34:07 -0800-) From: jbrankin@ntlworld.com (Jim Brankin)m" Subject: Re: Fibre Channel and SCS< Message-ID: <863f19d6.0303060138.c71d817@posting.google.com>   Kerry,    F   Are there any plans to use 802.11b or other wireless LAN products as a cluster interconnect?    - Jim          "Main, Kerry" <Kerry.Main@hp.com> wrote in message news:<BE56C50EA024184DAF48F0B9A47F5CF402660DEF@kaoexc01.americas.cpqcorp.net>...d > Rob, > H > Where I expect SCS over FC will play a role is as a backup path shouldH > both network switches (assuming dual switch config) become unavailable > for any reason.- > I > In the traditional CI Cluster world, if you lose both network switches,sH > then the users are dropped but all batch jobs continue and the cluster > as a whole is not impacted.3 > I > In the SAN cluster of today, if you lose both network switches then yourJ > are in big trouble as all cluster communications have been lost. With FCJ > as a backup path, the cluster would continue as in the CI cluster world. > J > This is a good message for those Customers who have Operations folks who2 > do not trust their Network group and vice versa. > 	 > Regardse >  > Kerry Main > Senior Consultant  > Hewlett-Packard (Canada) Co.# > Consulting & Integration Servicesp > Voice: 613-592-4660t > Fax   : 613-591-4477 > Email: kerryDOTmain@hpDOTcom/ >     (remove the DOT's and replace with "."'s)t! > OpenVMS DCL - the original .COMm >  >  > -----Original Message-----5 > From: Rob Young [mailto:young r@encompasserve.org]   > Sent: March 5, 2003 10:38 AM > To: Info-VAX@Mvb.Saic.Comm$ > Subject: Re: Fibre Channel and SCS >  > ? > In article <v6c3t366g2b083@corp.supernews.com>, "Alan Boyles"c& > <alan.boyles@mindspring.com> writes:J > > Currently Fiber Channel cannot be used as a cluster interconnect, but F > > I hear that's changing.  Gigabit E/net is now a supported Cluster * > > Interconnect and has been for a while. > >  > @ > 	SCS over fibre was presented to me as a "great thing to have"
 > but the @ > 	performance isn't stellar (I guess shouldn't be either as you > think A > 	about the protocols involved, how many layers?).  As mentioned 	 > before,*C > 	SCS over fibre for DT "mostly" and that is borne out on slide 11  > @ > 	of the latest roadmap and it is a wide box on that powerpoint > 	(form mid-2004 to 2006).  > 	 > 				Rob0   ------------------------------   Date: 6 Mar 2003 09:49:45 -0800T1 From: keithparris_NOSPAM@yahoo.com (Keith Parris)6" Subject: Re: Fibre Channel and SCS= Message-ID: <cf15391e.0303060949.348882cb@posting.google.com>f  h wmarsh@etrade.com (Wil Marsh) wrote in message news:<36f52bd6.0303051624.147326a6@posting.google.com>...H > Hopefully they will never put SCS on Fiberchannel. SCS traffic ideally6 > should be on dedicated paths throughout the cluster.  D But look at this from another viewpoint: Fibre Channel has Fast_PathA support today (which you were wishing for on LANs).  And if Fibrei= Channel supported SCS, what would prevent you from creating aaF dedicated (even 2-gigabit) Fibre Channel SAN for SCS just like you can? create a dedicated LAN for SCS today with Fast and/or 1-Gigabitl	 Ethernet?n  F Also, just because SCS on FC is supported doesn't mean you have to useB it.  You could always adjust the priority downward using SCACP (onF 7.3-1 or above) so that FC would only be used as a last resort, but itE could provide additional redundancy to keep the cluster together when C the network goes bad.  And occasionally the network does go bad, itr
 seems. ;-)   ------------------------------   Date: 6 Mar 2003 10:02:20 -0800n1 From: keithparris_NOSPAM@yahoo.com (Keith Parris)i" Subject: Re: Fibre Channel and SCS= Message-ID: <cf15391e.0303061002.1aa26e27@posting.google.com>w  R brandon@dalsemi.com wrote in message news:<03030606293955@dscis6-0.dalsemi.com>... > Will Marsh wrote:e1 > > We tried memory channel and that caused cpu 0-A > > overloads due to lack of fastpath support for memory channel.o > H > I have a small cluster (2 DS20 and 1 ES40) using MC-II; I am seriouslyO > considering replacing with a private 100 Mbit network - for similiar reasons,2O > well, maybe all conjecture on my part.  Anyone have objections, observations,  > thoughts on this?l  D First of all, I'd make sure you really are having the same problem.  Usei   $MONITOR MODES/CPU=0F to see how close your Primary CPU is to saturation in interrupt state.8  Or look at DECps interrupt-state data if you have that.  > If it turns out you do have this problem, it will not be untilB Fast_Path support for LANs is shipped that moving to a LAN for SCS would help you.   B (How many CPUs do you have in your DS20s and ES40?  Fast_Path only# helps for SMP machines, of course.)e  A In my measurements using Roy G. Davis' LOCKTIME* programs, MemoryeF Channel 2 had latencies of around 120 microseconds for lock requests. F LANs will have 1.5 to 2 times that latency, based on my measurements. C (If there's interest, I could make this tool available for folks toC use.)/   ------------------------------   Date: 6 Mar 2003 10:06:19 -0800E1 From: keithparris_NOSPAM@yahoo.com (Keith Parris)s" Subject: Re: Fibre Channel and SCS= Message-ID: <cf15391e.0303061006.501bfdcd@posting.google.com>k   rdeininger@mindspring.com (Robert Deininger) wrote in message news:<rdeininger-0603030816040001@user-uinj0ah.dialup.mindspring.com>...L > Unless you are out of PCI slots, you might as well keep the MC and add 100 > Mb networking.  Use both.e  F By default, SCS will prefer MC over Fast Ethernet, so just adding FastF Ethernet alone won't change things (other than providing redundancy toD ride through a MC failure).  With 7.3-1 or above, you can adjust the8 relative priorities of MC and Fast Ethernet using SCACP.   ------------------------------  % Date: Thu, 06 Mar 2003 12:11:51 -0600  From: brandon@dalsemi.comy" Subject: Re: Fibre Channel and SCS1 Message-ID: <03030612115167@dscis6-0.dalsemi.com>a  F > First of all, I'd make sure you really are having the same problem.  > Use  >   $MONITOR MODES/CPU=0H > to see how close your Primary CPU is to saturation in interrupt state.: >  Or look at DECps interrupt-state data if you have that. > @ > If it turns out you do have this problem, it will not be untilD > Fast_Path support for LANs is shipped that moving to a LAN for SCS > would help you.R   MON/MOD/CPU=0 K On a busy system Interrupt State runs 10-30% falling around 15% on average.p This is on a DS20E  D > (How many CPUs do you have in your DS20s and ES40?  Fast_Path only% > helps for SMP machines, of course.),   #1 = DS20E 2x500 MHz #2 = DS20E 2x500 MHz+ #3 = ES40  2x667 MHz (? me thinks off hand)e  C > In my measurements using Roy G. Davis' LOCKTIME* programs, MemorypH > Channel 2 had latencies of around 120 microseconds for lock requests. H > LANs will have 1.5 to 2 times that latency, based on my measurements. E > (If there's interest, I could make this tool available for folks to  > use.)R  ' I would like to have the code.  Please!R       John Brandon VMS Systems Administrator. Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wkr 972.371.4003 fxe   ------------------------------  % Date: Thu, 06 Mar 2003 12:25:03 -0600o From: brandon@dalsemi.coma" Subject: Re: Fibre Channel and SCS1 Message-ID: <03030612250392@dscis6-0.dalsemi.com>   H > By default, SCS will prefer MC over Fast Ethernet, so just adding FastH > Ethernet alone won't change things (other than providing redundancy toF > ride through a MC failure).  With 7.3-1 or above, you can adjust the: > relative priorities of MC and Fast Ethernet using SCACP.  2 I understand that SCS would prefer MC over Fast-E.  C I am limited on PCI slots and will be needing to make some changes.e   My options:2A 1) Replace MC with Fast-E; then use both Fast-E and GIGe for SCS.l- 2) Keep MC; and use both MC and GIGe for SCS.a  M The Fast-E will be dedicated to SCS traffic.  Potentially console traffic fortE the HSG80 controllers.  So very little other traffic will be on here.1  6 I will look forward to V7.3-1 for relative priorities.       John Brandon VMS Systems Administrator  Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wke 972.371.4003 fxo   ------------------------------  $ Date: Thu, 6 Mar 2003 07:37:26 -0600' From: "ProNetworkPC" <free@support.com> 1 Subject: Free Support, All OSes, All Applicationso5 Message-ID: <b47j0r$5gm$10599@sun-news.laserlink.net>o   http://www.pronetworkpc.como   -- Posted by News Bulk Poster Unregistered version   ------------------------------  % Date: Thu, 06 Mar 2003 14:18:06 -0000n4 From: VAXMAN <vaxman@dec.killed.the.goldengoose.com>  Subject: HP broke my link to WIS/ Message-ID: <v6em4udaulv939@corp.supernews.com>   K Okay, I tried to reach the WIS service today and get redirected to some HP B$ site that has nothing to do with it.  ! Anybody know what the URL is now?@  F God I'm tired of going on snipe hunts every time this company changes  hands.   ------------------------------  % Date: Thu, 06 Mar 2003 08:47:31 -0600r From: brandon@dalsemi.com $ Subject: Re: HP broke my link to WIS1 Message-ID: <03030608473127@dscis6-0.dalsemi.com>   M > Okay, I tried to reach the WIS service today and get redirected to some HP n& > site that has nothing to do with it. > # > Anybody know what the URL is now?  > H > God I'm tired of going on snipe hunts every time this company changes  > hands.  % http://relay.support.compaq.com:9004/p   John Brandon VMS Systems Administrator) Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wks 972.371.4003 fxo   ------------------------------  % Date: Thu, 06 Mar 2003 09:46:33 -0500n& From: Ken Robinson <kenrbnsn1@rcn.com>$ Subject: Re: HP broke my link to WIS< Message-ID: <5.1.0.14.2.20030306094458.06c2bba0@pop.rcn.com>  & At 02:18 PM 3/6/2003 +0000, you wrote:  K >Okay, I tried to reach the WIS service today and get redirected to some HPt% >site that has nothing to do with it.  >n" >Anybody know what the URL is now? >-F >God I'm tired of going on snipe hunts every time this company changes >hands.u  - What URL did you use. This one worked for me VJ <http://relay.support.compaq.com:9004/~wis>, after winding my way through  all the screens...   Ken    ------------------------------  # Date: Thu, 06 Mar 2003 17:40:29 GMTt" From:   VAXman-  @SendSpamHere.ORG$ Subject: Re: HP broke my link to WIS0 Message-ID: <00A1C766.08491A50@SendSpamHere.ORG>  f In article <v6em4udaulv939@corp.supernews.com>, VAXMAN <vaxman@dec.killed.the.goldengoose.com> writes: > L >Okay, I tried to reach the WIS service today and get redirected to some HP % >site that has nothing to do with it.  > " >Anybody know what the URL is now? >oG >God I'm tired of going on snipe hunts every time this company changes   >hands.  >r  I I've had WIS bookmarked my my Netscape for years and that URL still works-H to this day.  In fact, I just tried it to verify that it still does work before replying here.a  % http://relay.service.digital.com:9004t   --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COMm             5   "Well my son, life is like a beanstalk, isn't it?" :   ------------------------------  " Date: Thu, 6 Mar 2003 14:19:07 GMT0 From: Donald McCabe <donald.p.mccabe@boeing.com>= Subject: Re: HP killing Golden Eggs- future of goose unknown. ) Message-ID: <3E6758DB.5040504@boeing.com>o  6 The Eggs are invaluable and definitely worth the Gold.  
 VAXVMS wrote: 9 > Did anybody else find this tool to be extremely useful?c > $ > I think y'all know what to do. :^) >  > I already am.  >  > ========================" > William W. Webb - EMS Operations+ > OpenVMS Systems Support - USPS DSSC AnnexF. > 4730 Hargrove Road, Raleigh, NC 27616-2874  @ > 919.325.7500 x4186  <FirstInitialLastNameAtEmailDotUSPSDotGov> >    ------------------------------  $ Date: Thu, 6 Mar 2003 11:52:57 -0500 From: koskaj@bender.come= Subject: Re: HP killing Golden Eggs- future of goose unknown.s; Message-ID: <03030611525719.522.12632946@alaxp3.bender.com>D   Aren't the eggs located at ...  . http://h18000.www1.hp.com/info/golden-eggs/#T5  ' They certainly have been useful for me.u   :) jck
 John Koska JKoska@nospam.bender.com  P From:	SMTP%"donald.p.mccabe@boeing.com"  "Donald McCabe"  6-MAR-2003 09:45:29.14 To:	Info-VAX@Mvb.Saic.Comc CC:	: Subj:	Re: HP killing Golden Eggs- future of goose unknown.  6 The Eggs are invaluable and definitely worth the Gold.  
 VAXVMS wrote:e9 > Did anybody else find this tool to be extremely useful?  > $ > I think y'all know what to do. :^) >  > I already am.  >  > ========================" > William W. Webb - EMS Operations+ > OpenVMS Systems Support - USPS DSSC Annex . > 4730 Hargrove Road, Raleigh, NC 27616-2874  @ > 919.325.7500 x4186  <FirstInitialLastNameAtEmailDotUSPSDotGov> >    ------------------------------  % Date: Thu, 06 Mar 2003 09:16:42 +0100 ' From: JOUKJ <joukj@hrem.stm.tudelft.nl>m( Subject: iconv and UTF-8 on OpenVMS7.3-1* Message-ID: <b4705b$mln$1@news.tudelft.nl>   Hi all,o  > I found the following in some README of an opensource program:      D         The native iconv on Compaq Tru64 doesn't contain support forB         (UTF-8, so you'll need to use GNU libiconv instead. (WhenF         (using GNU libiconv for GTK+, you'll need to use GNU libiconvE           (for GNU gettext as well.) This probably applies to relatedn$         (operating systems as well.  G Should I assume that on a OpenVMS7.3-1 with the latest realesed DECCRTLo# the UTF-8 support is not available?                             Joukc   ------------------------------  % Date: Thu, 06 Mar 2003 15:51:27 +0000 ' From: Andrew Harrison SUNUK Consultancyt8 Subject: Re: Linus trashes Itanium - he's scared of VMS!. Message-ID: <3E676E7F.9000201@nospamn.sun.com>   Bill Todd wrote:M > "Andrew Harrison SUNUK Consultancy" <Andrew_No.Harrison_No@nospamn.sun.com> ; > wrote in message news:3E65EBAF.9030805@nospamn.sun.com...l >    >>+ >>I wasn't refering to TPC-C but to SPECint= >>but you are right. >> >>77K TPM for a 4 CPU Xeon MP  >>87K TPM for a 4 way Itanium 2  >  > K > I'm pretty sure that there used to be 4-processor results for the 2.4 GHz N > Xeons with 512 KB of on-chip cache - said results being about 10% lower than2 > the 2 GHz/2 MB cache Xeon result you note above. >   6 The 77K TPM result is for the 2 GHz XeonMP. I would be5 suprised if there are 2.4 GHz results published sinces5 the fastest Xeon-MP is currently 2 GHz. But that alsou8 may explain why you saw a result and now cannot find it, it could have been withdrawn.S   >  >>Both with MS-SQLserver >>0 >>The most up to date Xeon number is only with a. >>1 P 2.4 GHz box and it turns in 18K TPM. One* >>would expect the 2.8/3 GHz systems to be	 >>faster.n >  > L > As noted above, the amount of on-chip cache seems to be more important forK > TPC-C than the clock speed (and IIRC the fastest Xeons still support only, > 512 KB on chip). >  > 3 >>But then who uses TPC-C as a performance measure.t >  >  > Most of the world except Sun.o >  > 3 >>Point is that on Integer micro benchmarks Xeon is 7 >>competitive against a similarly configured Itanium 2.k >  > I > Then you should have so qualified your earlier, more general statement.6 >   7 Well on SPECint Xeon-MP wins by a small margin and Xeon 8 by wins by 25% which is a rather larger margin, on TPC-C4 Itanium is 13% ahead, with it being a 64bit cpu than its processing perfromance.     G >>110.6 is the number HP still have up on their web site and I have notcB >>found any references that suggest that this number is incorrect, >  > J > Came up in a discussion on comp.arch a while ago.  Supposedly the HP WebN > site was changed to reflect both the original and the corrected numbers, but > that's just hearsay. >   ' It doesn't appear to have been changed.    > ...a >  > E >>>That's all very interesting, but has absolutely no bearing on youra >>
 > original > J >>>misstatement (which was that "Xeons cost Intel a lot less to build than >> > IPF, > K >>>*with a 25% of the die size in a given process*" [emphasis added in caset >> > youl > F >>>had forgotten:  your original quote is still right up there above]. >>>/ >> >>F >>Well no not quite, implimenting Itanium 2 in a 130 nanometer process, >>should decrease the die size to ~314 mm^2. >  > I > There's no need to guess in this matter:  Intel has already stated thatEM > Madison's die area is 374 mm^2.  That, however, includes 6 MB of on-chip L3 I > cache compared with McKinley's 3 MB:  since the cache consumes about 35oM > mm^2/MB, if McKinley were shrunk to 130 nm its die area would thus be aboutS > 269 mm^2.m >   9 I am guessing just as much as you however as the basis oflD my guess I used the die shrink ratios for Xeon/P4 as they moved fromA 180 to 130 nanometer the same start and end process as the one weaD are discussing. Without anything else to go on its easily as good asD trying to guess how much the cache shrinks relative to the core whenF you move from a larger process to a smaller process which is what you  seem to be trying to do.  B Cache is easier to shrink because its regular the rest of the core? isn't. Hence the ratio of metal on Itanium used by the cache tow metal used by the core.p    M > No, you were wrong in your estimate of how large McKinley would be in a 130 K > nm process.  If Paul DeMone's 140 mm^2 value for P4 is correct, then it'soG > just over 52% of that; even if the 131 mm^2 value you came up with iseK > correct, then it's still 48.7% - nothing like 25% and close enough to 50%uL > for this discussion (the 70% high end of the range reflected a Xeon with 2I > MB of on-chip cache:  assuming the same 35 mm^2/MB density, it would be ? > 184 - 192 mm^2 - 68.4% - 71.4% of McKinley's size in 130 nm)./ >   A Well as I said earlier chipgeek lists P4 as 131 in 130 nanomater.e  8 You also make the huge and risky assumption that say two4 die sizes with an area of say 131 mm^2 result in the5 same number of dies per wafer, they don't because nota  all dies have the same geometry.  7 You have also just assumed that the die size will scalet3 linearly with the process density, this is also notn	 the case.   5 Still its nice to know that we are all just guessing.e  7 My guess is based on the shrink ratio for P4/Xeon yoursu$ is a bit more difficult to work out.   Regardso Andrew Harrisonl   ------------------------------  $ Date: Thu, 6 Mar 2003 13:06:30 -0500* From: "Bill Todd" <billtodd@metrocast.net>8 Subject: Re: Linus trashes Itanium - he's scared of VMS!2 Message-ID: <VaGdnUFbO6G3E_qjXTWcpw@metrocast.net>  K "Andrew Harrison SUNUK Consultancy" <Andrew_No.Harrison_No@nospamn.sun.com>m9 wrote in message news:3E676E7F.9000201@nospamn.sun.com...n >b >M > Bill Todd wrote:' > > "Andrew Harrison SUNUK Consultancy" ' <Andrew_No.Harrison_No@nospamn.sun.com>h= > > wrote in message news:3E65EBAF.9030805@nospamn.sun.com...o   ...r  H > >>Well no not quite, implimenting Itanium 2 in a 130 nanometer process. > >>should decrease the die size to ~314 mm^2. > >. > >oK > > There's no need to guess in this matter:  Intel has already stated thateL > > Madison's die area is 374 mm^2.  That, however, includes 6 MB of on-chip L3K > > cache compared with McKinley's 3 MB:  since the cache consumes about 35aI > > mm^2/MB, if McKinley were shrunk to 130 nm its die area would thus beI abouto
 > > 269 mm^2.f > >i > # > I am guessing just as much as youm  I No, you're guessing (incorrectly), while I'm reporting Intel's statementss about its own chip.a    however as the basis ofF > my guess I used the die shrink ratios for Xeon/P4 as they moved fromC > 180 to 130 nanometer the same start and end process as the one webF > are discussing. Without anything else to go on its easily as good asF > trying to guess how much the cache shrinks relative to the core whenG > you move from a larger process to a smaller process which is what youS > seem to be trying to do.  H No, it is not.  Intel has stated the size of Madison, and it's 374 mm^2.K Intel has stated that 57% of its die area is consumed by its 6 MB of cache,c@ which works out to 35.5 mm^2/MB (I rounded this down to 35 to beI conservative, since the percentage is only given to 2 digits of precision0K anyway).  From those two figures, it's only a *very* small extrapolation totF the size that McKinley would be in 130 nm:  266 - 269 mm^2 - in markedE contrast to your guess based on a completely different product with ay7 completely different cache/non-cache ratio on the chip.d   ...0  : > You also make the huge and risky assumption that say two6 > die sizes with an area of say 131 mm^2 result in the7 > same number of dies per wafer, they don't because notn" > all dies have the same geometry.  K You're either very confused or spinning again.  While you've been trying tofJ muddy the waters by changing the discussion to yields, I've said *nothing*D about that, just corrected your erroneous assertions about die size.   ...   7 > Still its nice to know that we are all just guessing.C  K I suppose it would be easier for you to think that than that you might just , be clueless.  But, once again, you're wrong.   - bill   ------------------------------   Date: 5 Mar 2003 23:16:50 -0800b# From: dooleys@snowy.net.au (dooley)oO Subject: Re: Maximum Record Size Error, -RMS-F-MRS, invalid maximum record sizel< Message-ID: <1ca82fc6.0303052316.419da2a@posting.google.com>  l del.duncan@intel.com (Del Duncan) wrote in message news:<98342c9f.0303051553.7c5e2e9c@posting.google.com>...D > I'm trying to use an FDL to create and convert files but I get theG > above error, even when creating a simple test file.    I'm attempting F > to create the files as Indexed, variable-length records, record size > 25042. > B > In debugging this, I find that if I set the FILE-ORGANIZATION toF > SEQUENTIAL, it then works, but if I set it to INDEXED, it fails.   IH > did some experimentation and found that I could get it to create a newA > file at 6500 but not at 7500.    The "OpenVMS Record ManagementDF > Utilities Reference Manual" states that Indexed files can have appr. > 32K records. > 3 I've just tried it and a record size of 16384 workso6 while a size of 32768 fails. (on alpha OpenVMS V7.2-1)" ...many versions of test.fdl later' it seems that the cutoff point is 32232a+ are your rms sysgen parameters different toe SYSGEN>  SHOW RMS  mN Parameter Name         Current    Default     Min.      Max.     Unit  DynamicN --------------         -------    -------    -------   -------   ----  -------L RMS_DFMBC                    16         16         1        127 Blocks     DL RMS_DFMBFSDK                  0          0         0        127 Blocks     DL RMS_DFMBFSMT                  0          0         0        127 Blocks     DL RMS_DFMBFSUR                  0          0         0        127 Buffers    DL RMS_DFMBFREL                  0          0         0        127 Buffers    DL RMS_DFMBFIDX                  0          0         0        127 Buffers    DL RMS_DFMBFHSH                  0          0         0        127 Buffers    DL RMS_PROLOGUE                  0          0         0          3 Prolog-Lvl DL RMS_EXTEND_SIZE               0          0         0      65535 Blocks     DK RMS_FILEPROT              64000      64000         0      65535 Prot-mask  GL RMS_DFNBC                     8          8         1        127 Blocks     DL RMS_DFLRL                     0          0         0      32767 Bytes      DL RMS_HEURISTIC                 0          0         0          1 Boolean    DL RMS_VCC_DFW                   0          0         0          1 Boolean    DL RMSD1                         0          0         0          1            D	 SYSGEN>      Phil   ------------------------------   Date: 6 Mar 2003 10:03:03 -0800 - From: jodonnell@hrblock.com (Jason O'Donnell)fO Subject: Re: Maximum Record Size Error, -RMS-F-MRS, invalid maximum record sizet= Message-ID: <9059bf6b.0303060656.25d86022@posting.google.com>t  / dooleys@snowy.net.au (dooley) wrote in message b5 > I've just tried it and a record size of 16384 worksa8 > while a size of 32768 fails. (on alpha OpenVMS V7.2-1)$ > ...many versions of test.fdl later) > it seems that the cutoff point is 32232C- > are your rms sysgen parameters different to   4 I would be curious to know if 32767 works for you...   ------------------------------   Date: 6 Mar 2003 10:06:45 -0800u- From: jodonnell@hrblock.com (Jason O'Donnell)s" Subject: Re: Move Over, Alpha Down= Message-ID: <9059bf6b.0303060723.7d045d4e@posting.google.com>f   eBay...o   ------------------------------  % Date: Thu, 06 Mar 2003 09:24:05 -0500 & From: David M Smith <dsmit115@csc.com> Subject: Re: MS Virus 8 Message-ID: <dcme6v8vthpvuiihhhiu06gqmpqc29c0pn@4ax.com>  G On Wed, 5 Mar 2003 06:39:03 -0800, "Tom Linden" <tom@kednos.com> wrote:e  E >There was a post the other day about a virus masquerading as a patchuI >from Microsoft.  Just got it here.  If I don't recognize mail, I read itt= >with vms mail, otherwise Outlook.  Here's what it looks liker >m >MAIL> read 205t( >    #205         5-MAR-2003 05:10:49.38 >MAIL , >From:   SMTP%"mzrllgg-fsqlkyfb@licqihq.com"% >To:     "MS Consumer"@hughestele.comi >CC: >Subj:   Internet Security PackI 	...D >Also, I am curious how to decipher this,  am I in a mailing list at >hughesteleb >called "MS Consumer"  ?  N I got the same email to the same email list (apparently). My internal security folks referred me to:o  I 	http://securityresponse.symantec.com/avcenter/venc/data/w32.gibe@mm.htmla  D to explain this particular virus. The attached file name on mine wasO "patch755.exe", which was stripped off by our corporate email filters (which dos, not allow .EXE attachments to pass through).I -------------------------------------------------------------------------bI David M. Smith 302.391.8533                       dsmit115 at csc dot comdI Computer Sciences Corporation     (Opinions are those of the writer only)tI -------------------------------------------------------------------------1   ------------------------------   Date: 6 Mar 2003 09:09:44 -0800c. From: kidehen@openlinksw.com (Kingsley Idehen)E Subject: Open Source Benchmark Utility for Oracle and other Databases = Message-ID: <caa49ee9.0303060518.10db130c@posting.google.com>,  l timasmith@hotmail.com (Tim Smith) wrote in message news:<a7234bb1.0303051508.4d2da069@posting.google.com>...r > Hein van den Heuvel <hein_netscape@eps.zko.dec.com> wrote in message news:<3E63C5E9.88569D3F@eps.zko.dec.com>... > > Tim Smith wrote: > > / > > > I see Oracle publishes some benchmarks ataI > > > http://www.oracle.com/apps_benchmark/, but notably there is nothingtK > > > for VMS - is that because Oracle writes to the filesystem, not directeL > > > to device file files directly i.e. VMS filesystem is a lot slower than > > > raw devices? > > 0 > > It has nothing to do with (potential) speed.D > > It is just a commercial / marketing decision mostly from Oracle.K > > They decided there was not enough critical mass to maintain support forcK > > the Oracle Applicaiton suite on VMS. The database itself is and will betM > > supported at one of the higher tier levels. Details are not up to me, buttL > > it used to be product release on VMS 90 days after first release. PleaseK > > verify with Oracle. They may also choose to skip 'dot' releases. Dunno,oN > > let's say they had 9.0, skipped 9.1 but released 9.2. Again, this is not aE > > statement of support, just a line of thinking. Check with Oracle.d > > L > > The VMS Filesystem is actually an advantage! The VMS filesystem does NOT > > buffer data.J > > On Unix systems the OS tends to waste time and memory buffering OracleL > > data pages which are better managerd by Oracle in its buffer pool (SGA).L > > On many Unix implementation, for ultimate Oracle perfromance once has toC > > deal with hard-to-manage 'Raw Devices' to avoid said buffering.cK > > On VMS you have the comfort of a file system for Alloaction, Naming andmK > > backups yet the speed of a raw device. On HP Tru64 Unix Oracle can (and-L > > will) use the DIRECT IO feature to get the same effect on single systems > > as well as in clusters.0 > > I > > > If anyone has older benchmarks that include VMS I would like to seee > > > them.t > > L > > It would be nice to see some VMS / Oracle benchmark, but I will not hold > > my breath.N > > Benchmarks require major investment which both companies believe is betterF > > spend on the products itself. VMS will offer comparable (ballpark)- > > performance as Unix on the same platform.sN > > It will not be 2x slower. It might be a little slower or a a little faster! > > depending on the application.dK > > The performance will be close enough to focus on other, more important, L > > platform decision factors: Cost-of-ownership, Reliability, Availability,/ > > Experience, Applications, Installed base,..i > > E > > Hope this helps some, but it is just an opinion from a guy in the- > > sidelines.M > > Be sure to contact officials at Oracle and HP for the official positions.  > >  > > Cheers,r
 > >     Hein.l > G > That makes sense, and I appreciate your comments.  The only problem IMG > have is that I find that simple I/O operations are much slower on VMSeH > than AIX.  I understand that this is a complex topic dependent on disk= > subsystems, disk speeds, etc etc but take this for example:y > E > I have two systems, one a Alpha 4100 with VMS 7.1 and a RS6000 H-50iC > both with 2GB RAM and 2 processors.  These systems were purchased D > around the same time for the same purpose and I would term them asH > comparable.  I did simple file copies from the same disk to same disk.G >  While I do not have the disk speeds, I used an older disk on the AIXnG > (4GB) and a newer disk on VMS (12GB).   I don't believe there was anyd > fragmentation to consider. > 7 > AIX: 297MB - 2MB/second, if repeated -  2.84MB/secondo > 9 > VMS: 297MB - 1.02MB/second, if repeated - 1.19MB/secondr > G > I don't mean to bash VMS, I just am trying to understand if it really F > comes close for I/O performance.  I've seen several places that haveF > to through a lot of hardware at performance bound applications.  AnyG > time I've done SQLPlus inserts they have been a lot faster on AIX.  IaE > know that many people still prefer VMS so I am trying to figure outg > why... >  > Tims  E There is a new Free and Open Source benchmark utility that allows youiA to perform objective industry standard benchmarks against any SQLo= database via ODBC or JDBC. It is available for download from:eA <http://www.openlinksw.com> . Note that you can perform real-time = comparative benchmarks side by side (Oracle vs SQL Server for ! instance) via a visual interface.    ------------------------------  % Date: Thu, 06 Mar 2003 08:39:43 -050002 From: rdeininger@mindspring.com (Robert Deininger)% Subject: Re: oracle benchmarks on VMSuL Message-ID: <rdeininger-0603030839440001@user-uinj0ah.dialup.mindspring.com>  = In article <a7234bb1.0303051508.4d2da069@posting.google.com>, ( timasmith@hotmail.com (Tim Smith) wrote:    F >That makes sense, and I appreciate your comments.  The only problem IF >have is that I find that simple I/O operations are much slower on VMSG >than AIX.  I understand that this is a complex topic dependent on disk < >subsystems, disk speeds, etc etc but take this for example: >sD >I have two systems, one a Alpha 4100 with VMS 7.1 and a RS6000 H-50B >both with 2GB RAM and 2 processors.  These systems were purchasedC >around the same time for the same purpose and I would term them aseG >comparable.  I did simple file copies from the same disk to same disk.nF > While I do not have the disk speeds, I used an older disk on the AIXF >(4GB) and a newer disk on VMS (12GB).   I don't believe there was any >fragmentation to consider.g >s6 >AIX: 297MB - 2MB/second, if repeated -  2.84MB/second >r8 >VMS: 297MB - 1.02MB/second, if repeated - 1.19MB/second  E I know nothing about the RS6000 hardware, and only a little about theoG AlphaServer 4100.  But your result leaves so many variables unspecifiednG that I think it says NOTHING about the capabilities of either operatingr system.p  B Details of the disks are important.  Alphaserver disks are usuallyI selected for performance, at least in part.  Aftermarket add-on disks arefF more often selected for low cost.  Your 12GB disk is an odd size for aF 4100, so I guess it's aftermarket.  It could easily be slower than the older, smaller disk.  H Were the disks on the same SCSI bus?  Were the SCSI adapters in the same PCI bus?  F The fact that both systems showed significant changes for the "repeat"C tests indicates that there is some caching happening somewhere.  Tol< account for that, you have to account for all the cache- and+ memory-related adjustments on both systems.r  H On VMS, there are many variables in how you do the I/O, and how you tuneC certain knobs in the process.  The defaults are very often far from J optimal, and this is a real and legitimate complaint against VMS.  RMS hasH some well-known bottlenecks.  But a little well-placed tuning can have a
 huge benefit.n  G And you might not have the defaults, you might have inherited somethingh even worse.a  : Similar comments may apply to AIX, but I dunno about that.  H Your VMS version is antiquated, and in some situations isn't a very good4 predictor of how current VMS (V7.3-1) would perform.    F >I don't mean to bash VMS, I just am trying to understand if it reallyE >comes close for I/O performance.  I've seen several places that havedE >to through a lot of hardware at performance bound applications.  AnyuF >time I've done SQLPlus inserts they have been a lot faster on AIX.  ID >know that many people still prefer VMS so I am trying to figure out >why...i  G The area where VMS shines is reliability.  For many, that is worth morey% than a moderate boost in performance.'  F It is good that you are trying to understand.  Tests of other people'sJ workloads aren't too helpful; it is your own workload that matters.  SpendJ a bit of time finding your bottlenecks.  There is a VMS performance manualE in the doc set.  If you find a bottleneck, you can usually get advicer5 (here and elsewhere) that will help you eliminate it.r  J For your disk-to-disk file copy on VMS, what where your process quotas and RMS parameters?    ------------------------------  % Date: Thu, 06 Mar 2003 15:15:42 +0000n' From: Andrew Harrison SUNUK Consultancy % Subject: Re: oracle benchmarks on VMSo. Message-ID: <3E67661E.5050304@nospamn.sun.com>   Robert Deininger wrote: ? > In article <a7234bb1.0303051508.4d2da069@posting.google.com>,s* > timasmith@hotmail.com (Tim Smith) wrote: >  >  > G >>That makes sense, and I appreciate your comments.  The only problem I G >>have is that I find that simple I/O operations are much slower on VMSoH >>than AIX.  I understand that this is a complex topic dependent on disk= >>subsystems, disk speeds, etc etc but take this for example:h >>E >>I have two systems, one a Alpha 4100 with VMS 7.1 and a RS6000 H-501C >>both with 2GB RAM and 2 processors.  These systems were purchasedWD >>around the same time for the same purpose and I would term them asH >>comparable.  I did simple file copies from the same disk to same disk.F >>While I do not have the disk speeds, I used an older disk on the AIXG >>(4GB) and a newer disk on VMS (12GB).   I don't believe there was anyu >>fragmentation to consider. >>7 >>AIX: 297MB - 2MB/second, if repeated -  2.84MB/seconda >>9 >>VMS: 297MB - 1.02MB/second, if repeated - 1.19MB/second) >  > G > I know nothing about the RS6000 hardware, and only a little about theyI > AlphaServer 4100.  But your result leaves so many variables unspecified I > that I think it says NOTHING about the capabilities of either operatingi	 > system.a >   =  From a CPU standpoint it depends on which Alpha CPU you havea9 in the 4100. The H50 has a 332 Mhz Power PC 604e CPU thato? does 14.4 SPECint95. The 4100 ranged from 7.5 to 18.8 SPECint95s depending the CPU used.g  . Like the 4100 the H50 supported up to 4 CPU's.  D > Details of the disks are important.  Alphaserver disks are usuallyK > selected for performance, at least in part.  Aftermarket add-on disks are3H > more often selected for low cost.  Your 12GB disk is an odd size for aH > 4100, so I guess it's aftermarket.  It could easily be slower than the > older, smaller disk. > J > Were the disks on the same SCSI bus?  Were the SCSI adapters in the same
 > PCI bus? > H > The fact that both systems showed significant changes for the "repeat"E > tests indicates that there is some caching happening somewhere.  To > > account for that, you have to account for all the cache- and- > memory-related adjustments on both systems.s > J > On VMS, there are many variables in how you do the I/O, and how you tuneE > certain knobs in the process.  The defaults are very often far fromtL > optimal, and this is a real and legitimate complaint against VMS.  RMS hasJ > some well-known bottlenecks.  But a little well-placed tuning can have a > huge benefit.p > I > And you might not have the defaults, you might have inherited somethingm
 > even worse.  > < > Similar comments may apply to AIX, but I dunno about that. > J > Your VMS version is antiquated, and in some situations isn't a very good6 > predictor of how current VMS (V7.3-1) would perform. >  >  > G >>I don't mean to bash VMS, I just am trying to understand if it reallyrF >>comes close for I/O performance.  I've seen several places that haveF >>to through a lot of hardware at performance bound applications.  AnyG >>time I've done SQLPlus inserts they have been a lot faster on AIX.  I:E >>know that many people still prefer VMS so I am trying to figure outo >>why... >  > I > The area where VMS shines is reliability.  For many, that is worth morei' > than a moderate boost in performance.u > H > It is good that you are trying to understand.  Tests of other people'sL > workloads aren't too helpful; it is your own workload that matters.  SpendL > a bit of time finding your bottlenecks.  There is a VMS performance manualG > in the doc set.  If you find a bottleneck, you can usually get advicet7 > (here and elsewhere) that will help you eliminate it.  > L > For your disk-to-disk file copy on VMS, what where your process quotas and > RMS parameters?1   ------------------------------  $ Date: Thu, 6 Mar 2003 11:38:44 -0500' From: "Main, Kerry" <Kerry.Main@hp.com>l% Subject: RE: oracle benchmarks on VMSsT Message-ID: <BE56C50EA024184DAF48F0B9A47F5CF402660DF5@kaoexc01.americas.cpqcorp.net>   Tim,  G As previous reply indicated, the OpenVMS OS version (7.1) is very old -E' probably about 6-7 years old right now..  % Is your version of AIX also this old?p  D As an example of recent improvements for OpenVMS and RMS, a previousH noter on comp.os.vms was able to achieve 80% better numbers by not using@ the default RMS and quota parameters and taking advantage of the following VMS V7.3-1 info:  E http://h71000.www7.hp.com/doc/731FINAL/6657/6657pro_007.html (see RMSpD New features which includes write-back capability that is similar to what UNIX systems handle IO)  E http://h71000.www7.hp.com/doc/731FINAL/6657/6657pro.html (See sectionT 1.1)   Regardsm  
 Kerry Main Senior Consultantf Hewlett-Packard (Canada) Co.! Consulting & Integration Services5 Voice: 613-592-4660n Fax   : 613-591-4477 Email: kerryDOTmain@hpDOTcom-     (remove the DOT's and replace with "."'s)t OpenVMS DCL - the original .COMa     -----Original Message-----1 From: Tim Smith [mailto:timasmith@hotmail.com]=20c Sent: March 5, 2003 6:08 PMi To: Info-VAX@Mvb.Saic.Com.% Subject: Re: oracle benchmarks on VMS-    D Hein van den Heuvel <hein_netscape@eps.zko.dec.com> wrote in message+ news:<3E63C5E9.88569D3F@eps.zko.dec.com>...g > Tim Smith wrote: >=200 > > I see Oracle publishes some benchmarks at=20J > > http://www.oracle.com/apps_benchmark/, but notably there is nothing=20E > > for VMS - is that because Oracle writes to the filesystem, not=20)H > > direct to device file files directly i.e. VMS filesystem is a lot=20 > > slower than raw devices? >=20. > It has nothing to do with (potential) speed.J > It is just a commercial / marketing decision mostly from Oracle. They=20H > decided there was not enough critical mass to maintain support for the  H > Oracle Applicaiton suite on VMS. The database itself is and will be=20J > supported at one of the higher tier levels. Details are not up to me,=20J > but it used to be product release on VMS 90 days after first release.=20B > Please verify with Oracle. They may also choose to skip 'dot'=20H > releases. Dunno, let's say they had 9.0, skipped 9.1 but released 9.2.  H > Again, this is not a statement of support, just a line of thinking.=20 > Check with Oracle. >=20I > The VMS Filesystem is actually an advantage! The VMS filesystem does=20lH > NOT buffer data. On Unix systems the OS tends to waste time and memory  H > buffering Oracle data pages which are better managerd by Oracle in its  H > buffer pool (SGA). On many Unix implementation, for ultimate Oracle=20F > perfromance once has to deal with hard-to-manage 'Raw Devices' to=20H > avoid said buffering. On VMS you have the comfort of a file system for  H > Alloaction, Naming and backups yet the speed of a raw device. On HP=20 > Tru64 Unix Oracle can (andE > will) use the DIRECT IO feature to get the same effect on single=20g! > systems as well as in clusters.  >=20J > > If anyone has older benchmarks that include VMS I would like to see=20	 > > them.  >=20H > It would be nice to see some VMS / Oracle benchmark, but I will not=20C > hold my breath. Benchmarks require major investment which both=20aG > companies believe is better spend on the products itself. VMS will=20 G > offer comparable (ballpark) performance as Unix on the same platform.oE > It will not be 2x slower. It might be a little slower or a a little  faster > depending on the application.:> > The performance will be close enough to focus on other, more
 important,< > platform decision factors: Cost-of-ownership, Reliability,
 Availability,-- > Experience, Applications, Installed base,..3 >=20F > Hope this helps some, but it is just an opinion from a guy in the=20E > sidelines. Be sure to contact officials at Oracle and HP for the=20  > official positions.+ >=20	 > Cheers,  >     Hein.i  E That makes sense, and I appreciate your comments.  The only problem IhE have is that I find that simple I/O operations are much slower on VMS F than AIX.  I understand that this is a complex topic dependent on disk; subsystems, disk speeds, etc etc but take this for example:V  H I have two systems, one a Alpha 4100 with VMS 7.1 and a RS6000 H-50 bothG with 2GB RAM and 2 processors.  These systems were purchased around thehF same time for the same purpose and I would term them as comparable.  IG did simple file copies from the same disk to same disk.  While I do notr5 have the disk speeds, I used an older disk on the AIX>E (4GB) and a newer disk on VMS (12GB).   I don't believe there was anyt fragmentation to consider.  5 AIX: 297MB - 2MB/second, if repeated -  2.84MB/secondt  7 VMS: 297MB - 1.02MB/second, if repeated - 1.19MB/second?  E I don't mean to bash VMS, I just am trying to understand if it reallygG comes close for I/O performance.  I've seen several places that have todF through a lot of hardware at performance bound applications.  Any timeE I've done SQLPlus inserts they have been a lot faster on AIX.  I knoweE that many people still prefer VMS so I am trying to figure out why...-   Tim    ------------------------------   Date: 6 Mar 2003 12:16:37 -0600e; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) 7 Subject: Re: OT Re: National Moratorium to Stop the War 3 Message-ID: <oBEVoU1GDarG@eisner.encompasserve.org>   U In article <3E669B69.9040000@MMaz.com>, "Barry Treahy, Jr." <Treahy@MMaz.com> writes:  >sJ > This is certainly not the place to be posting political 'opinions' that I > are, IMHO, totally lame and typically liberal (ie. no back bone except w5 > when its ok for Clinton to bomb aspirin factories!)=  H    You would like to see only those 'opinions' which are totaly lame and    typically conservative?      No.   ------------------------------   Date: 6 Mar 2003 09:23:52 -0800N( From: bob@instantwhip.com (Bob Ceculski)5 Subject: Re: OT:  National Moratorium to Stop the War-= Message-ID: <d7791aa1.0303060618.5c0673cd@posting.google.com>E  d "Bill Todd" <billtodd@metrocast.net> wrote in message news:<EJmcncf6AeXyRPujXTWcow@metrocast.net>...7 > "Bob Ceculski" <bob@instantwhip.com> wrote in messageg9 > news:d7791aa1.0303050546.688b2224@posting.google.com... 9 > > "Bill Todd" <billtodd@metrocast.net> wrote in messageP1 >  news:<Q5ednUHaAJy3NvijXTWc3A@metrocast.net>...AG > > > After last night's newscasts were full of descriptions of Dubya'sg
 >  increasinguL > > > rush to war (perhaps to start next week, according to some pundits), IK > > > decided to take a look around to see what people were doing about it.l >  TurnsF > > > out that today (March 5th) is the date of a planned 'moratorium' >  (letters toL > > > legislators, work stoppages, teach-ins, civil disobedience - the gamut >  of L > > > possible responses), and since the next major planned demonstration (a >  MarchK > > > 15th march on Washington) may turn out to be after the fact those who  >  wouldL > > > like to have a chance to register their opposition might want to do soG > > > today.  For further information, see www.internationalanswer.org.l > > >  > > > - bill > > : > > we have a nut in Iraq who wants to arm terrorists with: > > nuclear suitcase bombs which would make 9/11 look like5 > > a picnic, and you are on this vms board promotinga< > > flower power?  Didn't you learn anything from WWII Bill?; > > 9/11 was another Pearl Harbor, but you're still wanting  > > to appease murderers?n > I > Thank you, Bob, for making the quality of the opposing viewpoint clear.1K > Although those who agree with you might counter with the observation that , > even a stopped clock is right twice a day. >  > - bill  8 yes, history does repeat itself for people like you Bill# who don't learn a thing from it ...R   ------------------------------   Date: 6 Mar 2003 09:44:57 -0800.' From: eugene@cse.ucsc.edu (Eugene Miya)t! Subject: portability was/is: unixt& Message-ID: <3e678919$1@news.ucsc.edu>  H In article <b47ct6$ahj$11@bob.news.rcn.net>,  <jmfbahciv@aol.com> wrote:' >In article <3e6645a5$1@news.ucsc.edu>,P, >   eugene@cse.ucsc.edu (Eugene Miya) wrote:, >>In article <b45f4m$6he$0@205.153.154.199>,/ >>Patrick Scheible  <kkt@itchy.serv.net> wrote: G >>>While converting wouldn't be painless, it would be easier than, say,p" >>>converting from TOPS-10 to VMS. >>> >>That depends on how the application was coded under TOPS-10.I >>If the application programmer was trying to be too efficient (in memoryrF >>[packed characters] and speed) and had only experience on the DEC-10E >>architectures (some called that arrogance), that would be difficult B >>(I've had my share [SPICE] of 36-bit to 32-bit conversions which >>I care not to repeat). >>D >>If the application programmer was experienced in different machineI >>architectures, and coded for portability, a port might not be too hard.s >tD >Or if the app programmer RTFM, he could have written portable code.@ >We did document extensions to the standard.  IIRC, most of themA >had to do with I/O.  Some of those little OSes didn't believe ins >disks and directories ;-).     6 That's why I worded my comment to Patrick in that way.  A The temptation in that era was just too much for most programmersaG to pass up getting the fifth character into storage.  As programmers weoE got too cute packing characters (just look at MS/Word) because memorys was so expensive in those days.w  D I/O isn't the only problem (as another pointed out the I/O example).# 4 bits isn't trivial in some cases.t  E Could be worse.  Could have been 6 characters in a 36-but Univac word 2 or 10 characters in a 60-bit CDC word like Pascal.   ------------------------------  $ Date: Thu, 6 Mar 2003 13:07:52 -0500. From: "Leo Demers" <leo_dot_demers@HP_dot_com>% Subject: Re: Questions on SSH for VMS , Message-ID: <3e679082$1@usenet01.boi.hp.com>  L Folks this will be up on the webpage shortly put it can't hurt to have it in two places.      - Leol   INSTALLATION NOTE:  C Corrections for Installing hp TCP/IP Services for OpenVMS SSHv2 EAKn   Description:  < After installing the SSHv2 EAK, perform the following steps:  H 1. The SSHv2 EAK kit fails to extract the updated templates text library4 (TCPIP$TEMPLATES.TLB) from the kit. To correct this:  D a. Extract the updated template text library from the SSHv2 EAK kit:  @ $ PRODUCT EXTRACT FILE TCPIP_SSH /SELECT=TCPIP$TEMPLATES.TLB_SSH  # b. Copy the templates text library:r  E $ COPY TCPIP$TEMPLATES.TLB_SSH SYS$COMMON:[SYSLIB]TCPIP$TEMPLATES.TLBa      L 2. The updated TCPIP$UCP.EXE image must be installed as a known image beforeE any SSHv2 EAK components can be configured using the TCPIP$CONFIG.COMm command procedure.  G To install the updated TCPIP$UCP.EXE image, choose one of the following  methods:  8  On systems where network availability is not critical:  ! $ @SYS$STARTUP:TCPIP$SHUTDOWN.COMP  $ @SYS$STARTUP:TCPIP$STARTUP.COM   or  4  On systems where network availability is critical:  2 $ INSTALL REPLACE SYS$COMMON:[SYSEXE]TCPIP$UCP.EXE2 $ INSTALL REPLACE SYS$COMMON:[SYSMSG]TCPIP$MSG.EXE     --
 Leo Demers  OpenVMS Security Product Manager Leo_dot_Demers_at_HP_dot_COM& <dittman@dittman.net> wrote in message1 news:kmt9a.27992$ES3.3388@nwrddc04.gnilink.net...h! > Dirk Munk <munk@home.nl> wrote:eF > > If the PCSI kit is the same as we got this morning, then please be advised thatE > > there are some small errors in the installation and configuration  procedures.s > > 3 > > In the kit is a new version of the text libraryi  sys$library:TCPIP$TEMPLATES.TLB,J > > it is named TCPIP$TEMPLATES.TLB_SSH. The installation procedure should	 copy thisgK > > file as a new version to sys$library, but does not do so. You will havef toL > > extract the file from the PCSI kit by hand, and place it in sys$library. Do thisn# > > before trying to configure SSH.- >.I > To clarify, extract TCPIP$TEMPLATES.TLB_SSH and place it in SYS$LIBRARYi renamedt > to TCPIP$TEMPLATES.TLB >eL > > The configuration routine of the server also contains a small error. The service H > > SSH is defined (TCPIP SET SERVICE etc.) with the option /FLAG=TCPIP. This option F > > is invalid. Define the service manually without this option before configuringl$ > > the SSH server, and all is well. >iB > I removed the /flag=tcpip from TCPIP$CONFIG.COM and that worked. > -- > Eric Dittman > dittman@dittman.nets   ------------------------------  % Date: Thu, 06 Mar 2003 09:40:05 -0500a+ From: Steve Lionel <Steve.Lionel@intel.com>-% Subject: Re: Recent posts I have made08 Message-ID: <gdne6voorlt5pkg543k63ul7rrsu4qqo1k@4ax.com>  > On 5 Mar 2003 07:16:29 -0800, susan_skonetski@hotmail.com (Sue Skonetski) wrote:X  F >Just a reminder that the information that I am posting on cov may not' >have been circulated to all of hp yet.a   Well, they have now! :-)   Stevec   ------------------------------  % Date: Thu, 06 Mar 2003 13:04:44 +0000 @ From: Martin hepworth <martinh@iosolid-spammerstate-nologic.com>0 Subject: restoring a file from a backup (newbie)0 Message-ID: <3e67476e$1_1@nnrp1.news.uk.psi.net>   Hi guysm  G well first time for everyhting I guess. Trying to restore a nuked file   from tape backup - vms 7.0.    the backup script does this...=   BACKUP/IMAGE/noassist/LABEL="''LABEL'"/REWIND/BLOCK=32768 - 8          /RECORD/IGNORE=(NOBACKUP,INTERLOCK)/FAST DKA0:  MKA500:BACKUP.BCK /SAVEa    L how do I get a specifi file back (eg SYS$SPECIFIC:[UCX_LPD]UCX$PRINTCAP.DAT)   -- Martin3 (remove 'io spammer no' words to get my email addr)    ------------------------------  % Date: Thu, 06 Mar 2003 12:35:33 -0600i From: brandon@dalsemi.comu4 Subject: Re: restoring a file from a backup (newbie)1 Message-ID: <03030612353358@dscis6-0.dalsemi.com>   	 > Hi guys  > I > well first time for everyhting I guess. Trying to restore a nuked file s > from tape backup - vms 7.0.g >   > the backup script does this...? >   BACKUP/IMAGE/noassist/LABEL="''LABEL'"/REWIND/BLOCK=32768 -o: >          /RECORD/IGNORE=(NOBACKUP,INTERLOCK)/FAST DKA0:  > MKA500:BACKUP.BCK /SAVEn >  > N > how do I get a specifi file back (eg SYS$SPECIFIC:[UCX_LPD]UCX$PRINTCAP.DAT) >  > Martin     Write lock the tape - I hope...    $ create/dir dka0:[restore]oD $ backup MKA500:BACKUP.BCK /save /select=(myfile,...) dka0:[restore]   This should get you started.   John Brandon VMS Systems Administratorr Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wk  972.371.4003 fxi   ------------------------------  % Date: Thu, 06 Mar 2003 13:38:54 -0500s& From: Ken Robinson <kenrbnsn1@rcn.com>4 Subject: Re: restoring a file from a backup (newbie)< Message-ID: <5.1.0.14.2.20030306133351.070a7988@pop.rcn.com>  & At 01:04 PM 3/6/2003 +0000, you wrote:   >Hi guys > H >well first time for everyhting I guess. Trying to restore a nuked file  >from tape backup - vms 7.0. >e >the backup script does this...e> >  BACKUP/IMAGE/noassist/LABEL="''LABEL'"/REWIND/BLOCK=32768 -K >         /RECORD/IGNORE=(NOBACKUP,INTERLOCK)/FAST DKA0: MKA500:BACKUP.BCK u > /SAVEt >y > M >how do I get a specifi file back (eg SYS$SPECIFIC:[UCX_LPD]UCX$PRINTCAP.DAT)e  L $back/rew/sel=(sys$specific:[ucx_lpd]ucx$printcap.dat) mka500:backup.bac/sav  C Note: you really don't need the "/sav" on a backup saveset on tape.    Kenl   ------------------------------  % Date: Thu, 06 Mar 2003 12:42:06 -0600m From: brandon@dalsemi.coml4 Subject: Re: restoring a file from a backup (newbie)1 Message-ID: <03030612420697@dscis6-0.dalsemi.com>2   > N > $back/rew/sel=(sys$specific:[ucx_lpd]ucx$printcap.dat) mka500:backup.bac/sav > E > Note: you really don't need the "/sav" on a backup saveset on tape.  >  > Kens    8 $back/rew/sel=(sys$specific:[ucx_lpd]ucx$printcap.dat) -  mka500:backup.bac/sav -  dka0:  6 But you really do need the destination drive... *grin*     John Brandon VMS Systems Administrator  Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wkG 972.371.4003 fxE   ------------------------------  $ Date: Thu, 6 Mar 2003 11:58:42 -0000 From: "John" <john@dateline.gg>a Subject: RMS version numbern( Message-ID: <b47d1r$cjb$1@ns2.cwgsy.net>  J Can anybody tell me how to get the version number of a newly created file?   ------------------------------   Date: 6 Mar 2003 06:22:10 -0600r- From: Kilgallen@SpamCop.net (Larry Kilgallen)6 Subject: Re: RMS version number)3 Message-ID: <dm1+oTHKQWqQ@eisner.encompasserve.org>h  J In article <b47d1r$cjb$1@ns2.cwgsy.net>, "John" <john@dateline.gg> writes:L > Can anybody tell me how to get the version number of a newly created file?  G Terse questions lead us to make assumptions.  You could post again with- clarification.  G I presume you would not have asked if looking for the latest version ofs the file would work...  I ...so you must be in a situation where multiple versions of the same file F are being created, and you want the version created by _your_ process.  I So on the VMS V7.3 (most recent version for your VAX) documentation CDROMR0 that came with your copy of the operating system  @ 	file:///VMSDOC073/v73/4523/4523pro_021.html#creat_nam_in_fields  H is a table indicating that NAM$B_RSL is updated and the buffer specifiedE by NAM$L_RSA is updated after your call to $CREATE.  That string will0F contain the full file specification of the file you just created.  Use) the $FILESCAN system service described atM  4 	file:///VMSDOC073/v73/4527/4527pro_040.html#jun_250   to extract the version number.   ------------------------------  % Date: Thu, 06 Mar 2003 06:40:36 -0600  From: brandon@dalsemi.com  Subject: Re: RMS version number 1 Message-ID: <03030606403617@dscis6-0.dalsemi.com>   L > Can anybody tell me how to get the version number of a newly created file?  8 $ version = f$integer(f$parse(filespec,,,"VERSION")-";")   John Brandon VMS Systems AdministratorD Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wk  972.371.4003 fx    ------------------------------  % Date: Thu, 06 Mar 2003 10:08:42 -0500m' From: Chris Olive <nospam@raytheon.com>a Subject: Re: RMS version number9; Message-ID: <7xJ9a.92$35.700@dfw-service2.ext.raytheon.com>n   Larry Kilgallen wrote:L > In article <b47d1r$cjb$1@ns2.cwgsy.net>, "John" <john@dateline.gg> writes: > L >>Can anybody tell me how to get the version number of a newly created file? >  > I > Terse questions lead us to make assumptions.  You could post again withd > clarification. > I > I presume you would not have asked if looking for the latest version of- > the file would work... > K > ...so you must be in a situation where multiple versions of the same file H > are being created, and you want the version created by _your_ process. > K > So on the VMS V7.3 (most recent version for your VAX) documentation CDROM:2 > that came with your copy of the operating system > B > 	file:///VMSDOC073/v73/4523/4523pro_021.html#creat_nam_in_fields > J > is a table indicating that NAM$B_RSL is updated and the buffer specifiedG > by NAM$L_RSA is updated after your call to $CREATE.  That string will H > contain the full file specification of the file you just created.  Use+ > the $FILESCAN system service described at  > 6 > 	file:///VMSDOC073/v73/4527/4527pro_040.html#jun_250 >   > to extract the version number.  I Or assuming some FAB context is available (either because you've already tF got a FAB context/pointer from working code, or you get a FAB context I from say LIB$FIND_FILE for instance), then chain to the NAM from the FAB  H and use NAM$B_VER and NAM$L_VER to get the version number (string) from  the ESA.  F Somehow, I have a feeling this question springs from something deeper I than command line access for which the answer is a bit more obvious (and r? has already been offered.)  Maybe a continuation of a question  C concerning COPYing from a program in the last day or two... or not.E   Chrisn -----a Chris Olive  Systems Consultant' Raytheon Technical Services Corporation- Indianapolis, IN  * email: olivec(AT)indy(DOT)raytheon(DOT)com   ------------------------------  $ Date: Thu, 6 Mar 2003 15:17:22 -0000 From: "John" <john@dateline.gg>- Subject: Re: RMS version number-( Message-ID: <b47omb$7cu$1@ns2.cwgsy.net>  * "John" <john@dateline.gg> wrote in message" news:b47d1r$cjb$1@ns2.cwgsy.net...L > Can anybody tell me how to get the version number of a newly created file? >n  K Sorry, I was too brief in the original question - high pressure moment madem me forget myself.m  L I am inside a BASIC program and have just opened a file for writing.  I wantJ to know what version of the file I have just created, as there may alreadyJ be versions in existence, and other versions may follow later; yet I stillI want to be able to refer back to this particular one.  In the past I haveoJ only ever had to refer back to the latest version, but a change in the way) we do things has necessitated the change.e  J My alternative would be to abandon the existing file naming convention andI replace it with something more sensible - but there are many years of thetI existing structure in place, and "not starting from here" isn't really anm= easy option.  (But is starting to look more attractive......)o   ------------------------------  % Date: Thu, 06 Mar 2003 10:00:23 -0600s From: brandon@dalsemi.com  Subject: Re: RMS version numbert1 Message-ID: <03030610002339@dscis6-0.dalsemi.com>u  N > I am inside a BASIC program and have just opened a file for writing.  I wantL > to know what version of the file I have just created, as there may alreadyL > be versions in existence, and other versions may follow later; yet I stillK > want to be able to refer back to this particular one.  In the past I haveML > only ever had to refer back to the latest version, but a change in the way+ > we do things has necessitated the change.a > L > My alternative would be to abandon the existing file naming convention andK > replace it with something more sensible - but there are many years of thetK > existing structure in place, and "not starting from here" isn't really ant? > easy option.  (But is starting to look more attractive......)l    < There are system service (?) and RTL calls that you can use.  
 LIB$FILE_SCAN 
 LIB$FIND_FILEr  L However, if I understand what you are attempting to do, then this may not be enough.e  O You will need to concern yourself with multiple versions of the same file beingnM created at once.  If this is the case, then the RTL calls may fetch the wrong  file - my assumption.m  O I would follow the non-version method.  Use a date-time stamp when creating theh file:h   myfile_yyyymmddhhmmsstt.txt      John Brandon VMS Systems Administratorr Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wko 972.371.4003 fx    ------------------------------  % Date: Thu, 06 Mar 2003 11:19:38 -0500e' From: Chris Olive <nospam@raytheon.com>r Subject: Re: RMS version numberu; Message-ID: <CzK9a.93$35.703@dfw-service2.ext.raytheon.com>C   John wrote:p, > "John" <john@dateline.gg> wrote in message$ > news:b47d1r$cjb$1@ns2.cwgsy.net... > L >>Can anybody tell me how to get the version number of a newly created file? >> >  > M > Sorry, I was too brief in the original question - high pressure moment made  > me forget myself.  > N > I am inside a BASIC program and have just opened a file for writing.  I wantL > to know what version of the file I have just created, as there may alreadyL > be versions in existence, and other versions may follow later; yet I stillK > want to be able to refer back to this particular one.  In the past I havesL > only ever had to refer back to the latest version, but a change in the way+ > we do things has necessitated the change.e > L > My alternative would be to abandon the existing file naming convention andK > replace it with something more sensible - but there are many years of thedK > existing structure in place, and "not starting from here" isn't really ane? > easy option.  (But is starting to look more attractive......)e >   A Okay, yeah that helps.  So to refer back to my earlier post that BD indicated FABs and NAMs could be used, you may want to explorer the F USEROPEN option on the BASIC OPEN statement to accomplish this.  From 8 the documentation on USEROPEN for Compaq BASIC, we have:   <INDENT>B "BASIC calls the user program after it fills the FAB (File Access H Block), the RAB (Record Access Block), and the XABs (Extended Attribute H Blocks). The subprogram must issue the appropriate RMS calls, including B $OPEN and $CONNECT, and return the RMS status as the value of the F function. See the Compaq BASIC for OpenVMS Alpha and VAX Systems User 8 Manual for more information about the USEROPEN routine."	 </INDENT>s  G The filled FAB should have a NAM hanging off of it that you can use to  H get the version number from using the information I supplied in my last F post (NAM$B_VER holds the length of the version string in the ESA and H NAM$L_VER is the longword address pointing to the version number string " in the ESA with the ";" included.)  E So, I would think the USEROPEN function (which you have to write and 0G supply to your OPEN clauses, which I assume you already have), is your .: answer.  It may sound a bit involved, but it's really not.  G Haven't written any BASIC under VMS since, oh... 1987?  So, I'm loathe nH to offer more detail as it relates to BASIC itself.  Still, I know this > isn't a hard thing to do (to write your own USEROPEN routine.)   Chrisn -----s Chris Olivem Systems Consultant' Raytheon Technical Services Corporationv Indianapolis, IN  * email: olivec(AT)indy(DOT)raytheon(DOT)com   ------------------------------  $ Date: Thu, 6 Mar 2003 16:24:46 -0000 From: "John" <john@dateline.gg>o Subject: Re: RMS version numberm( Message-ID: <b47skm$gvj$1@ns2.cwgsy.net>  & <brandon@dalsemi.com> wrote in message+ news:03030610002339@dscis6-0.dalsemi.com... K > > I am inside a BASIC program and have just opened a file for writing.  Is wantF > > to know what version of the file I have just created, as there may already/H > > be versions in existence, and other versions may follow later; yet I stilltH > > want to be able to refer back to this particular one.  In the past I haveJ > > only ever had to refer back to the latest version, but a change in the wayI- > > we do things has necessitated the change.  > >mJ > > My alternative would be to abandon the existing file naming convention anddI > > replace it with something more sensible - but there are many years oft thelJ > > existing structure in place, and "not starting from here" isn't really anA > > easy option.  (But is starting to look more attractive......)s >  >e> > There are system service (?) and RTL calls that you can use. >a > LIB$FILE_SCANh > LIB$FIND_FILEs >eK > However, if I understand what you are attempting to do, then this may not  be	 > enough.p >iK > You will need to concern yourself with multiple versions of the same filea beingoI > created at once.  If this is the case, then the RTL calls may fetch thes wrong  > file - my assumption.o >uD > I would follow the non-version method.  Use a date-time stamp when creating the > file:s >r > myfile_yyyymmddhhmmsstt.txta  L I had been hoping to find a means whereby I could pass the channel number ofH the opened file to something, and for it to return me the full filespec.I $FILESCAN seems to me to need to be passed the full filespec in the firsttL place.  It seemed like such a simple thing a couple of hours ago - hence theG brevity of my original post.  It's a pain, because there is information K within the name of the file, that it is a shame to lose, (such as the id of J the creator of the file) but if I have to create another file to act as an  index to that, then so be it....  # Thanks for your assistance however.u   ------------------------------  % Date: Thu, 06 Mar 2003 11:36:53 -0500k' From: Chris Olive <nospam@raytheon.com>I Subject: Re: RMS version numberi; Message-ID: <NPK9a.94$35.647@dfw-service2.ext.raytheon.com>    John wrote: , > "John" <john@dateline.gg> wrote in message$ > news:b47d1r$cjb$1@ns2.cwgsy.net... > L >>Can anybody tell me how to get the version number of a newly created file? >> >  > M > Sorry, I was too brief in the original question - high pressure moment made  > me forget myself.t > N > I am inside a BASIC program and have just opened a file for writing.  I wantL > to know what version of the file I have just created, as there may alreadyL > be versions in existence, and other versions may follow later; yet I stillK > want to be able to refer back to this particular one.  In the past I havesL > only ever had to refer back to the latest version, but a change in the way+ > we do things has necessitated the change.  > L > My alternative would be to abandon the existing file naming convention andK > replace it with something more sensible - but there are many years of theeK > existing structure in place, and "not starting from here" isn't really anp? > easy option.  (But is starting to look more attractive......)t >   F Hey, hey...  The USEROPEN routine you could use is almost written for % you in the Compaq BASIC User's Guide:e  F http://h71000.www7.hp.com/doc/73final/cobol/bas_um_023.htm#index_x_896  F Certainly a better and more clear example that I could have provided. 3 Massage it a bit and you're in business, me thinks.s   Chrisr -----I Chris Olivei Systems Consultant' Raytheon Technical Services Corporationh Indianapolis, IN  * email: olivec(AT)indy(DOT)raytheon(DOT)com   ------------------------------  $ Date: Thu, 6 Mar 2003 09:09:55 -0800$ From: Shane Smith <ssmith@icius.com> Subject: RE: RMS version numberB0 Message-ID: <01C2E3C0.34E53DE0@sulfer.icius.com>  . F$SEARCH() followed by F$PARSE() should do it.   ShaneR   -----Original Message-----$ From: John [mailto:john@dateline.gg]& Sent: Thursday, March 06, 2003 3:59 AM To: Info-VAX@Mvb.Saic.Comk Subject: RMS version numberp    D Can anybody tell me how to get the version number of a newly created file?d   ------------------------------  % Date: Thu, 06 Mar 2003 18:41:20 +0100 ' From: Bitnissen <bitnissen@hotmail.com>4 Subject: Re: RMS version numbern8 Message-ID: <bi1f6vct4ka11odaf4p943mp01dn4f0d42@4ax.com>   Hi Johnt  ( I get the file version this way (pascal)  . function find_revision (var fab    : fab$type;.                         var rab    : rab$type;6                         var f      : ftype) : integer;   beginB  version := '';i  _status:=$CREATE(fab);e   if odd(_status) then      _status  := $connect(rab)   else6       writeln('NEWPRTFHR-E-CRE, error during create :'*               ,_status,'/',fab.fab$l_stv);  find_revision := _status;  $display(fab);o%  nam_ptr ::unsigned := fab.fab$l_NAM;"/  version_ptr :: unsigned := nam_ptr^.nam$l_ver;p6  version := substr(version_ptr^,1,nam_ptr^.nam$b_ver);  end;   / prt_name := 'kfk$spool:' + x_prt_name + '.lis';s_ open(prt,prt_name,history:=new,record_length:=2048,user_action:=Find_revision,error:=continue);e3 prt_name := prt_trim(prt_name) + prt_trim(version);e   m.v.h.   Jimmi       C On Thu, 6 Mar 2003 16:24:46 -0000, "John" <john@dateline.gg> wrote:s   >o' ><brandon@dalsemi.com> wrote in messagew, >news:03030610002339@dscis6-0.dalsemi.com...L >> > I am inside a BASIC program and have just opened a file for writing.  I >wantvG >> > to know what version of the file I have just created, as there maya >alreadyI >> > be versions in existence, and other versions may follow later; yet I  >stillI >> > want to be able to refer back to this particular one.  In the past I  >haveSK >> > only ever had to refer back to the latest version, but a change in thea >way. >> > we do things has necessitated the change. >> >K >> > My alternative would be to abandon the existing file naming conventions >andJ >> > replace it with something more sensible - but there are many years of >theK >> > existing structure in place, and "not starting from here" isn't reallye >anrB >> > easy option.  (But is starting to look more attractive......) >> >>? >> There are system service (?) and RTL calls that you can use.s >> >> LIB$FILE_SCAN >> LIB$FIND_FILE >>L >> However, if I understand what you are attempting to do, then this may not >ber
 >> enough. >>L >> You will need to concern yourself with multiple versions of the same file >beingJ >> created at once.  If this is the case, then the RTL calls may fetch the >wrong >> file - my assumption. >>E >> I would follow the non-version method.  Use a date-time stamp when-
 >creating the- >> file: >> >> myfile_yyyymmddhhmmsstt.txt >5M >I had been hoping to find a means whereby I could pass the channel number of I >the opened file to something, and for it to return me the full filespec.aJ >$FILESCAN seems to me to need to be passed the full filespec in the firstM >place.  It seemed like such a simple thing a couple of hours ago - hence thesH >brevity of my original post.  It's a pain, because there is informationL >within the name of the file, that it is a shame to lose, (such as the id ofK >the creator of the file) but if I have to create another file to act as ang! >index to that, then so be it....: > $ >Thanks for your assistance however. >5   ------------------------------  % Date: Thu, 06 Mar 2003 13:41:34 -0500l' From: Chris Olive <nospam@raytheon.com>  Subject: Re: RMS version numberh; Message-ID: <GEM9a.95$35.828@dfw-service2.ext.raytheon.com>e   Shane Smith wrote:0 > F$SEARCH() followed by F$PARSE() should do it. >  > ShaneI >  > -----Original Message-----& > From: John [mailto:john@dateline.gg]( > Sent: Thursday, March 06, 2003 3:59 AM > To: Info-VAX@Mvb.Saic.Comp > Subject: RMS version number  >  > F > Can anybody tell me how to get the version number of a newly created > file?  >  >   F So what happens if someone else or some other process creates another F file of the same name inbetween the time John opens the file and does E the $SEARCH and $PARSE?  He's going to get the version number of the r other process's open file...  ? Also F$SEARCH and F$PARSE are DCL.  John needs this from a 3GL.h  F I say a file scan of any kind, if I understand John's problem, leaves F the door open for finding another version of the same file that could I have been created in the time between when he opens HIS file and when he  H would scan for it (be it ANY of the file scanning suggestions made thus / far either using RMS routines or RTL routines.)   H If John can say "No one else will be creating a file at the same time I @ might be," then fine, maybe a file scan is a way to do it.  But G otherwise I say the door is open to getting version number that is NOT tE the same as the file he just opened.  The USEROPEN routine overcomes t this scenario.   Chrisr -----h Chris Olive  Systems Consultant' Raytheon Technical Services Corporationy Indianapolis, IN  * email: olivec(AT)indy(DOT)raytheon(DOT)com   ------------------------------  $ Date: Thu, 6 Mar 2003 10:42:50 -0800$ From: Shane Smith <ssmith@icius.com> Subject: RE: RMS version numbers0 Message-ID: <01C2E3CD.3326D920@sulfer.icius.com>  H Short question, short answer... ;-) Most of the information there wasn't available when I responded.l  = I agree, a useropen may be the correct approach, based on the G information now available. Someone already posted a link to an example.a   Shane'   -----Original Message-----. From: Chris Olive [mailto:nospam@raytheon.com]' Sent: Thursday, March 06, 2003 10:42 AMt To: Info-VAX@Mvb.Saic.Com  Subject: Re: RMS version number      Shane Smith wrote:0 > F$SEARCH() followed by F$PARSE() should do it. >  > Shane  >  > -----Original Message-----& > From: John [mailto:john@dateline.gg]( > Sent: Thursday, March 06, 2003 3:59 AM > To: Info-VAX@Mvb.Saic.Comt > Subject: RMS version numberr >  > F > Can anybody tell me how to get the version number of a newly created > file?a >  >   F So what happens if someone else or some other process creates another F file of the same name inbetween the time John opens the file and does E the $SEARCH and $PARSE?  He's going to get the version number of the t other process's open file...  ? Also F$SEARCH and F$PARSE are DCL.  John needs this from a 3GL.o  F I say a file scan of any kind, if I understand John's problem, leaves F the door open for finding another version of the same file that could H have been created in the time between when he opens HIS file and when heH would scan for it (be it ANY of the file scanning suggestions made thus / far either using RMS routines or RTL routines.)n  H If John can say "No one else will be creating a file at the same time I @ might be," then fine, maybe a file scan is a way to do it.  But G otherwise I say the door is open to getting version number that is NOT  E the same as the file he just opened.  The USEROPEN routine overcomes   this scenario.   Chrisl -----0 Chris Olive  Systems Consultant' Raytheon Technical Services Corporation  Indianapolis, IN  * email: olivec(AT)indy(DOT)raytheon(DOT)com   ------------------------------  * Date: Thu, 6 Mar 2003 15:37:38 +0000 (UTC)- From: lewis@spyder.mitre.org (Keith A. Lewis)d< Subject: Re: SCSI cluster disk thrashing between connections. Message-ID: <b47q02$3uc$1@newslocal.mitre.org>   brooks@cuebid.zko.dec.nospam (Rob Brooks) writes in article <0fBsBz+GJZPm@cuebid.zko.dec.com> dated Thu, 06 Mar 2003 02:15:40 GMT:E >The thing is that the other nodes on the shared bus should also haveeE >problems with that device, so it's somewhat surprising that at leastaB >one node is able to communicate with no problems.  It could pointD >to a problem with the HBA on MAZDA::, or a cabling problem, etc ...  J The problem only occurs when both hosts are attached to the SCSI bus.  TheL terminators are on the cables (with Y-adapters), and I checked the adapters'L SCSI IDs -- they are 14 and 15.  So I don't think there's a hardware problemJ with the cable or interface.  It's a fairly long cable but well within theE spec for differential SCSI.  The one "unsupported" thing I'm doing is0J running a KZPSA in a DS20E.  It almost doesn't fit because of stuff in the; way inside the box, luckily the end PCI/EISA slot was free.a  B >If you are going to MSCP-serve disks for any length of time, makeA >sure that any relevant SYSGEN params are tuned.  MSCP_BUFFERS is J >especially important on any serving system.  The default of 1024 (I think7 >that's the default as of 7.3-1) may not be sufficient.-  L I'll keep that in mind.  This cluster lives on MSCP.  Every node serves someJ disks, mounts all of them, and uses most.  1024 is what MSCP_BUFFER is setI to now -- that's way up from historical values in older AGEN$PARAM.REPORTdE files.  Is there some minimum you'd recommend, such as 1% of physicalt memory?   + --Keith Lewis              klewis$mitre.org > The above may not (yet) represent the opinions of my employer.   ------------------------------  # Date: Thu, 06 Mar 2003 16:45:45 GMT:/ From: brooks@cuebid.zko.dec.nospam (Rob Brooks) < Subject: Re: SCSI cluster disk thrashing between connections- Message-ID: <w3GfOrjRVct+@cuebid.zko.dec.com>s  / lewis@spyder.mitre.org (Keith A. Lewis) writes:  > brooks@cuebid.zko.dec.nospam (Rob Brooks) writes in article <0fBsBz+GJZPm@cuebid.zko.dec.com> dated Thu, 06 Mar 2003 02:15:40 GMT: > C >>If you are going to MSCP-serve disks for any length of time, makeeH >>sure that any relevant SYSGEN params are tuned.  MSCP_BUFFERS [sic] isK >>especially important on any serving system.  The default of 1024 (I thinkh8 >>that's the default as of 7.3-1) may not be sufficient. > N > I'll keep that in mind.  This cluster lives on MSCP.  Every node serves someL > disks, mounts all of them, and uses most.  1024 is what MSCP_BUFFER is setK > to now -- that's way up from historical values in older AGEN$PARAM.REPORTaG > files.  Is there some minimum you'd recommend, such as 1% of physical 	 > memory?   G Well, the V7.3-1 default is 1024 (up from 128, I think), so independent G of whatever autogen would calculate, 1024 would be as low as it can be.t  M $ MONITOR MSCP is a useful tool to see if you need to up MSCP_BUFFER.  If theiN Buffer Wait Rate is 0, then you're in good shape.  Any time that we're waitingM for a free buffer will be noted in the buffer wait rate.  I really don't have K any good suggestions for tuning this, however;  I've tended to do it by the  seat of my pants.   K Also, you can disable the multipath autofailback mechanism (the moving of a3H device from MSCP to a local path when a local path becomes available) byI setting the SYSGEN parameter MPDEV_AFB_INTVL to 0.  So, you can start out<E on a local path and jump to the MSCP path whenever the local ones getlL toasted, then stay on the remote from then on.  This is a dynamic parameter.   -- t  M Rob Brooks    VMS Engineering -- I/O Exec Group     brooks!cuebid.zko.dec.como   ------------------------------  ! Date: Thu, 06 Mar 03 11:09:26 GMT, From: jmfbahciv@aol.comm Subject: Re: unix>, Message-ID: <b47ct6$ahj$11@bob.news.rcn.net>  & In article <3e6645a5$1@news.ucsc.edu>,+    eugene@cse.ucsc.edu (Eugene Miya) wrote:m+ >In article <b45f4m$6he$0@205.153.154.199>,h. >Patrick Scheible  <kkt@itchy.serv.net> wrote:F >>While converting wouldn't be painless, it would be easier than, say,! >>converting from TOPS-10 to VMS.c >e= >That depends on how the application was coded under TOPS-10.nH >If the application programmer was trying to be too efficient (in memoryE >[packed characters] and speed) and had only experience on the DEC-10pD >architectures (some called that arrogance), that would be difficultA >(I've had my share [SPICE] of 36-bit to 32-bit conversions which/ >I care not to repeat).O > C >If the application programmer was experienced in different machinerH >architectures, and coded for portability, a port might not be too hard.  C Or if the app programmer RTFM, he could have written portable code.t? We did document extensions to the standard.  IIRC, most of themt@ had to do with I/O.  Some of those little OSes didn't believe in disks and directories ;-).   <snip>   /BAH    ' Subtract a hundred and four for e-mail.i   ------------------------------  ! Date: Thu, 06 Mar 03 11:12:44 GMTa From: jmfbahciv@aol.comw Subject: Re: unixe, Message-ID: <b47d3b$ahj$12@bob.news.rcn.net>  * In article <b45f4m$6he$0@205.153.154.199>,/    Patrick Scheible <kkt@itchy.serv.net> wrote:e >jmfbahciv@aol.com writes: >Q >t >> [about Unix]M >>D >> On the contrary, this may be its Achilles' heel.  There is no oneC >> entity to babysit the bits, enforce standards, and have the last C >> say on Bad Ideas.  Do not misunderstand me.  Given a choice, I'd/D >> always vote for the side that distributes the knowledge; however,@ >> there's a complete set of other problems that comes with that	 >> choice  >rF >There's not one entity for all of Unix, but each of the BSDs and each= >of the Linux distributions has a group babysitting its bits.b >yG >Open source gives some protection to users against arbitrary decisionsPE >by companies.  Multiple open source versions of Unix gives even morePE >protection.  Good ideas can get copied, while organizational failureT@ >by one group still leaves other versions of Unix around.  While? >converting wouldn't be painless, it would be easier than, say,P  >converting from TOPS-10 to VMS.  ; Sigh!  I understand this.  None of these people will sign a ; contract with an interested user site to guarantee support.t= That's why DEC could do a lot of business.  They had hundredsS= of able and smart people who could be thrown at _one_ problema: if the manpower was warrantied.  Running open source code = doesn't guarantee that my business' time-critical applicationM8 will have somebody out there dropping everything to work' until my computing problem is resolved.n   /BAH  ' Subtract a hundred and four for e-mail.3   ------------------------------  ! Date: Thu, 06 Mar 03 10:59:13 GMT  From: jmfbahciv@aol.comm Subject: Re: unixr+ Message-ID: <b47ca1$ahj$9@bob.news.rcn.net>   ) In article <w1Uf0EQpJryo@elias.decus.ch>, -    p_sture@elias.decus.ch (Paul Sture) wrote:kF >In article <b3sofh$rp0$2@bob.news.rcn.net>, jmfbahciv@aol.com writes:, >> In article <0PL33tx$iQG4@elias.decus.ch>,0 >>    p_sture@elias.decus.ch (Paul Sture) wrote:H >>>In article <b3qhii$nu2$4@bob.news.rcn.net>, jmfbahciv@aol.com writes: >>> 	 >>><snip>h >>>eC >>>> Unix could suck eggs but people will still use it if, and only ? >>>> if, they get the sources.  One of DEC's fatal mistakes wasI? >>>> to try to keep the "knowledge", a.k.a. sources, to itself;T> >>>> it was under the misconception that secrecy would protectA >>>> its investment and intellectual property.  This is one case (& >>>> where the exact opposite is true. >>>> C >>> B >>>Memories of a DEC comms program back in the late 1970s here. ItD >>>could have saved us an awful lot of money but didn't work. We hadD >>>the source for at least some of it, but all the comments had been >>>stripped. Plain stupid IMO. >>>tA >> I don't believe that.  I would believe that the comments never3
 >> existed.  a >> m >tE >This was a piece of RT11 macro which was supposed to talk to an IBM.m >0E >It appeared that someone had run an editing macro against the source2> >to search for a semicolon and then delete to the end of line. >oG >I don't remember the exact details, but the evidence convinced me thaty >was what had happened.e  C The only reason (that I can think of) is that the IBM protocol used-@ to talk to the device was proprietary(sp?) and the last edit to A the source had to be stripping the documenation (which is rather j: silly but then those lawyer-genrated edicts usually were).   /BAH  ' Subtract a hundred and four for e-mail.e   ------------------------------  ! Date: Thu, 06 Mar 03 11:01:35 GMT  From: jmfbahciv@aol.com0 Subject: Re: unixr, Message-ID: <b47cee$ahj$10@bob.news.rcn.net>  2 In article <1046869304.495878@saucer.planet.gong>,H    "Rupert Pigott" <roo@dark-try-removing-this-boong.demon.co.uk> wrote:; >"Brian 'Jarai' Chase" <bdc@world.std.com> wrote in messageI$ >news:b55hj-50s.ln1@pez.jarai.com... >[SNIP]o >oI >> This is true.  I think with VMS it's quite a lot more work because of 0 how H >> VMS was implemented, historically.  They've got to port compilers forJ >> BLISS, MACRO-32, C, ADA, and (others?) because those languages have allJ >> been used to implement various bits of the OS. In following comp.os.vmsJ >> recently, my understanding is that they're actually compiling VAX MACROK >> code into their target platforms of Alpha of IA64.  I bet that's got to 6 be > B >I very much doubt that VAX MACRO poses a huge challenge. You haveD >a fairly well known original target (the VAX architecture), whereas@ >if you're porting stuff like C across you're fighting more than@ >one layer... ie : The ISA, the compiler, and the OS. Although I7 >would bet that the VMS sources are fairly tidy by now.   A I wouldn't bet ;-).  If the IA64 is going to MACRO, it's the besta news I've heard.     /BAH  ' Subtract a hundred and four for e-mail.    ------------------------------  $ Date: Thu, 6 Mar 2003 14:16:37 -0000D From: "Rupert Pigott" <roo@dark-try-removing-this-boong.demon.co.uk> Subject: Re: unix 2 Message-ID: <1046960197.849918@saucer.planet.gong>  K <jmfbahciv@aol.com> wrote in message news:b47cee$ahj$10@bob.news.rcn.net...H   [SNIP]  C > I wouldn't bet ;-).  If the IA64 is going to MACRO, it's the bestl > news I've heard.  8 Oddly the VAX & IA-64 may have a lot in common, although6 the IA-64's political impact is far larger from what I	 can tell.R   Cheers,E Rupert   ------------------------------   Date: 6 Mar 2003 16:58:09 GMTw( From: bill@cs.uofs.edu (Bill Gunshannon) Subject: Re: unixR6 Message-ID: <b47un1$1so515$1@ID-135708.news.dfncis.de>  7 In article <20030306171513.62af6ea1.steveo@eircom.net>,n/ 	Steve O'Hara-Smith <steveo@eircom.net> writes:   > On Thu, 06 Mar 03 11:12:44 GMT > jmfbahciv@aol.com wrote: >  > JC> Running open source code aC > JC> doesn't guarantee that my business' time-critical applications> > JC> will have somebody out there dropping everything to work- > JC> until my computing problem is resolved.m > A > 	Nope, you have to pay for that kind of promise - and even then H > it is not often delivered IME. The good news is that there are outfits1 > offering payed support for open source systems.   aD Sometimes paying doesn't get it either (and I'm not talking about MSB here!)  I have held jobs where my purpose was to provide just suchE support for commercial, proprietary systems because turning in an SPR-C wasn't guaranteed to result in a fix at all, much less a quick one.<@ (I always remember a complaint made about something a particularH vendors FORTRAN compiler did that was fixed by adding an warning message9 that advised against doing that particualr operation. :-)-  F The advantage to Open Source is at least you are assured of having the8 source if you choose to have the problem fixed yourself.   bill   -- @J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>       ------------------------------  $ Date: Thu, 6 Mar 2003 17:05:56 -0000D From: "Rupert Pigott" <roo@dark-try-removing-this-boong.demon.co.uk> Subject: Re: unixa2 Message-ID: <1046970356.424884@saucer.planet.gong>  G "Rupert Pigott" <roo@dark-try-removing-this-boong.demon.co.uk> wrote in 4 message news:1046960197.849918@saucer.planet.gong...& > <jmfbahciv@aol.com> wrote in message& news:b47cee$ahj$10@bob.news.rcn.net... >0 > [SNIP] >5E > > I wouldn't bet ;-).  If the IA64 is going to MACRO, it's the bestR > > news I've heard. >f: > Oddly the VAX & IA-64 may have a lot in common, although   Eeek ! I should qualify that...h  < + VAX was "the future" and signalled the end of the line for* a fondly remembered architecture, the -10.< + They are both single-sourced architectures (ie : you won't= get AMD making IA-64 clones because Intel wouldn't let them).a7 + They are both fairly unique and extreme examples of am particular style of ISA.= + They are both "high-end" parts - by that I mean not "cheap"m, (certainly the case when compared to SPARC).  3 Apologies for the heart failures I may have caused.n   Cheers,s Rupert   ------------------------------   Date: 6 Mar 2003 11:24:15 -0600l; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler)  Subject: Re: unix 3 Message-ID: <8Sp2iWKTI3Mj@eisner.encompasserve.org>0  y In article <1046960197.849918@saucer.planet.gong>, "Rupert Pigott" <roo@dark-try-removing-this-boong.demon.co.uk> writes:d > : > Oddly the VAX & IA-64 may have a lot in common, although8 > the IA-64's political impact is far larger from what I > can tell.   D    Never underestimate the impact of the VAX.  For over a decade theE    rule of thumb was:  if you had to do a computer based system, you o=    went out and bought a VAX.  Even IBM and Microsoft did it.l  G    Sadly, the VAX was also instrumental to the evolution and popularityrD    of UNIX.  Of those folks at Berkely hadn't ported it to a VAX the$    rest of might never have seen it!   ------------------------------  $ Date: Thu, 6 Mar 2003 17:15:13 +0100, From: Steve O'Hara-Smith <steveo@eircom.net> Subject: Re: unixh7 Message-ID: <20030306171513.62af6ea1.steveo@eircom.net>l   On Thu, 06 Mar 03 11:12:44 GMT jmfbahciv@aol.com wrote:   JC> Running open source code tA JC> doesn't guarantee that my business' time-critical applicationp< JC> will have somebody out there dropping everything to work+ JC> until my computing problem is resolved.   ? 	Nope, you have to pay for that kind of promise - and even then F it is not often delivered IME. The good news is that there are outfits/ offering payed support for open source systems.n   --  D C:>WIN                                      |     Directable MirrorsJ The computer obeys and wins.                |A Better Way To Focus The SunH You lose and Bill collects.                 |  licenses available - see:F                                             |   http://www.sohara.org/   ------------------------------  * Date: Thu, 6 Mar 2003 13:48:17 +0000 (UTC)- From: lewis@spyder.mitre.org (Keith A. Lewis)a8 Subject: Re: Upgrade to OPEN VMS 7.3-1 - Patches needed?. Message-ID: <b47jj1$35e$1@newslocal.mitre.org>   norm.raphael@metso.com writes in article <OF77724E4B.9D92BBF0-ON85256CE0.0075E2A4@metso.com> dated Wed, 5 Mar 2003 16:58:30 -0500: >p >The upgrade checklist says: >(F >       If you use AGEN$INCLUDE in SYS$SYSTEM:MODPARAMS.DAT to includeH >        file(s) containing additional parameter settings, and the filesG >        which are being included are not on the system disk, the filesnF >        should be moved to the system disk and the AGEN$INCLUDE entry- >        updated to reflect the new location.  >a9 >which certainly seems to indicate you did nothing wrong.r  K The include file is physically on the system disk, but the path to it needs J a logical which I define in the custom section of SYLOGICALS.COM, so maybeI the upgrade procedure couldn't find it.  At this point I don't think it'snH worth a formal complaint; I'm not going to be upgrading any more systems@ with non-default DEVICE_NAMING or SHADOW_SYS_DISK parameters.  5B conversational boots is less work than writing a complaint anyway.  I >Since you cannot upgrade a shadowed system disk, I am not sure about theb, >value of shadow_sys_disk after the upgrade.  G I did mount a member as unshadowed while the upgrade CD was booted, butoI setting up the cluster to boot that way would have been several times thelK amount of work.  I have done the upgrades this way since 6.2-1H2 or so, and-/ this is the first time I have had this problem.0  + --Keith Lewis              klewis$mitre.org > The above may not (yet) represent the opinions of my employer.   ------------------------------   Date: 6 Mar 2003 09:58:15 -0800p' From: eugene@cse.ucsc.edu (Eugene Miya)o Subject: VAX again: unix& Message-ID: <3e678c37$1@news.ucsc.edu>  3 In article <8Sp2iWKTI3Mj@eisner.encompasserve.org>,f: Bob Koehler <koehler@eisner.aspm.encompasserve.org> wrote:. >   Never underestimate the impact of the VAX.   True.aF However for leverage, the PDP-11, the PDP-8, and the PDP-1 on one hand? and the PDP-6/PDP-10/DEC-20 had greater leverage in their time.bH This became sad upon seeing the VAX 9000.  The interaction and real time@ process control on the minicomputers (and midis) was what really distanced DEC from IBM.d   >    For over a decade theF >   rule of thumb was:  if you had to do a computer based system, you > >   went out and bought a VAX.  Even IBM and Microsoft did it.  > While I know they, and Apple and Cray too, owned VAXen, it was+ and continues to be an IBM dominated world.   H >   Sadly, the VAX was also instrumental to the evolution and popularityE >   of UNIX.  Of those folks at Berkely hadn't ported it to a VAX thee% >   rest of might never have seen it!i  E Gee, others thought it was the failure of various DEC code word namede+ architectures to keep track of performance.i  H I can see it now ... disgruntled VMS users build a cyborg Terminator andB send it back in time to bump off a young Bill Joy or Bob Fabry....   ------------------------------  + Date: Thu, 06 Mar 2003 08:16:26 +0100 (MET)09 From: Phillip Helbig <HELBPHI@sysdev.deutsche-boerse.com>:: Subject: Re: VAXstation 4000 VLC VMS 5.5-2 Boot Parameters; Message-ID: <01KT748C4L2A9FQH4P@sysdev.deutsche-boerse.com>n  E > I have an old 4000 VLC running VMS 5.5-2 and would like to have thenF > machine boot up without benefit of keyboard or mouse and then accessF > it via telnet and nfs over the lan.  right now it needs the keyboardC > plugged in and i have to type "boot" when i power it up. once ucxgG > comes up i can telnet it to it.  does anybody know the boot parameteriH > settings necessary to do this?  the system admin documentation has not > been seen for a while.  F Since you mention "mouse", I assume that OPA0 is the graphics monitor.G You can switch this to a serial terminal by flipping the "S3" switch (ItA think; I have some 4000s but not a VLC).  To avoid having to typesC "boot", you need to specify another value of "halt" at the console i prompt.e   ------------------------------  % Date: Thu, 06 Mar 2003 08:13:51 +0000r( From: Nic Clews <sendspamhere@127.0.0.1>: Subject: Re: VAXstation 4000 VLC VMS 5.5-2 Boot Parameters) Message-ID: <3E67033F.C32E0B45@127.0.0.1>m   Phillip Helbig wrote:  > G > > I have an old 4000 VLC running VMS 5.5-2 and would like to have theaH > > machine boot up without benefit of keyboard or mouse and then accessH > > it via telnet and nfs over the lan.  right now it needs the keyboardE > > plugged in and i have to type "boot" when i power it up. once ucxtI > > comes up i can telnet it to it.  does anybody know the boot parameterrJ > > settings necessary to do this?  the system admin documentation has not > > been seen for a while. > H > Since you mention "mouse", I assume that OPA0 is the graphics monitor.I > You can switch this to a serial terminal by flipping the "S3" switch (IsC > think; I have some 4000s but not a VLC).  To avoid having to typeND > "boot", you need to specify another value of "halt" at the console	 > prompt.   D And it's worth mentioning the three possible values for this type of. machine, and the implications of each setting.  5 BOOT, HALT and RESTART are the three notional values.r   HALT is self explanatory.u  E BOOT will, on successful power on, try to boot the default device and G flags as set at the console, and in the absence of anything attempts touH boot over the first ethernet adapter it finds. If the system crashes, itD will remain halted after the crash dump file is written (if it can).  F RESTART is similar to boot, the difference is, after a crash and dump, it will then auto reboot.   A On a micro class VAX you use the console command SET HALT <3,2,1>uB On an Alpha generally you set the console variable SET AUTO_ACTION <HALT,BOOT,RESTART>o  % (VAX 1 = restart, 2 = boot, 3 = halt)i -- T? Regards, Nic Clews a.k.a. Mr. CP Charges, CSC Computer Sciencesa nclews at csc dot com2   ------------------------------  % Date: Thu, 06 Mar 2003 06:31:46 -0600e From: brandon@dalsemi.com$: Subject: Re: VAXstation 4000 VLC VMS 5.5-2 Boot Parameters1 Message-ID: <03030606314626@dscis6-0.dalsemi.com>   H > Since you mention "mouse", I assume that OPA0 is the graphics monitor.I > You can switch this to a serial terminal by flipping the "S3" switch (IlC > think; I have some 4000s but not a VLC).  To avoid having to typeaD > "boot", you need to specify another value of "halt" at the console	 > prompt.     K I have both 4000 and 4000 VLC, I am curious about that too.  I guess I willa$ look at my VLC when I get to work...   John Brandon VMS Systems Administratort Dallas Semiconductor john.brandon@dalsemi.com 972.371.4172 wk  972.371.4003 fx    ------------------------------  % Date: Thu, 06 Mar 2003 15:07:33 +0000x' From: Andrew Harrison SUNUK Consultancyi! Subject: Re: VMS Backup solutions , Message-ID: <3E676435.70002@nospamn.sun.com>   Bob Koehler wrote:Z > In article <3E6625D5.4040107@nospamn.sun.com>, Andrew Harrison SUNUK Consultancy writes: >  > C >>>   Guess again.  I HAVE restored VMS backups from damaged tapes,  >>>   no data loss.e >>>n >>8 >>Well of course, but that depends on whats on the tape. >  > D >    You still don't get it.  The reason I could restore data from aJ >    damaged tape was because it had a VMS BACKUP.  VMS BACKUP is designed >    to deal with this.    >   5 Sorry you don't get it the tape in question was jamed 6 if you think that VMS BACKUP would have got you out of9 this problem then please explain how it physically unjams  a tape cartridge.s  6 And the Veritas Netbackup was able to recover the data7 because it keeps two copies on two separate tapes whichi9 would also be the only way that OpenVMS backup could havec recovered the same data.   Regardst Andrew Harrisonp   ------------------------------   Date: 6 Mar 2003 11:14:43 -0600t; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) ! Subject: Re: VMS Backup solutionst3 Message-ID: <Rts5XNg+EvQp@eisner.encompasserve.org>e  c In article <pP20T2VhKPwP@eisner.encompasserve.org>, Kilgallen@SpamCop.net (Larry Kilgallen) writes:0c > In article <841c6v04k6e31li53a76gu8p7snbgis1r7@4ax.com>, jlsue <jlsuexxxz@screaminet.com> writes:i > G >> And, in response to Larry's note about other OpenVMS disk with alias A >> entries.... OUCH!  I've never needed to implement them for any@F >> application I've managed, but I can empathize with the problem that >> will present. > J > But have you ever managed a system and implemented a method to _prevent_ > users from creating aliases ?   B    I've never seen any of the captive accounts I've used manage toC    create an alias or anything else I didn't explicitly want it to.t   ------------------------------   Date: 6 Mar 2003 11:19:35 -0600 ; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler)r! Subject: Re: VMS Backup solutionsI3 Message-ID: <oN1JqyAn0gUW@eisner.encompasserve.org>t  V In article <3E676435.70002@nospamn.sun.com>, Andrew Harrison SUNUK Consultancy writes:  7 > Sorry you don't get it the tape in question was jamedn8 > if you think that VMS BACKUP would have got you out of; > this problem then please explain how it physically unjamse > a tape cartridge.t  B    Nope, I've never done that with a cartridge.  Of couse I did doF    it with a jammed 9-track, and the cartridges I had to unjam myself.  8 > And the Veritas Netbackup was able to recover the data9 > because it keeps two copies on two separate tapes whicho; > would also be the only way that OpenVMS backup could have  > recovered the same data.  F    Nope.  I didn't say VMS BACKUP could recover data from a completelyC    destroyed tape.  The point is that I haven't found a native UNIXrG    backup utility which could recover data from a ratty tape, which VMSn    BACKUP does.-  D    I said something to the effect that I'm not happy with UNX backupG    utilities and I'm still not.  I shouldn't have to buy extra software-#    to accomplish such a basic task.o  9    OBTW, what does Veritas do on a one tape drive system?R   ------------------------------  # Date: Thu, 06 Mar 2003 15:19:40 GMTt3 From: "James T. Sprinkle" <oicmrsnakes@hotmail.com>S Subject: Re: VMS Question???0 Message-ID: <gGJ9a.15$Si.3256@news-west.eli.net>  L Thanks!  I actually have an emulator for the VAX that came bundled in with aK PDP-11 emulator.  Didn't think of that.  DUH!  I asked about the DECstation4J because I know of where I can pick one (or several) up for about $20 each.K But if it doesn't run VMS, I am not interested.  I have my RS/6000's for myu UNIX of choice (AIX).a   JaYmZ:  A "Tim Llewellyn" <tim.llewellyn@blueyonder.co.uk> wrote in message ) news:3E6673BA.B639238@blueyonder.co.uk...; >  >i > "James T. Sprinkle" wrote: > >EH > > Hi!  I worked with VMS in college several years back on both VAX and Alpha G > > architecture.  I am considering getting a hobby copy of VMS and was'K > > wondering if it would work on a DecStation?  Also, anybody know where IuI > > could look for used VAX/Alpha hardware at a reasonable price?  Thanksp forr
 > > any help!v > >n	 > > JaYmZ_ >^H > VMS does not run on the DECStation line with MIPS processors, if thats what you mean. >iJ > You can run the simh VAX emulator on Windows or unix if you just want to3 > play with VMS and are happy with the VAX version.e > 
 > regards, >  > --  > tim.llewellyn@blueyonder.co.uk >"J > * PLEASE NOTE tim.llewellyn@cableinet.co.uk address is NO LONGER VALID *   ------------------------------   Date: 6 Mar 2003 11:29:19 -0600a; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler)w Subject: Re: VMS Question???3 Message-ID: <w16rEUoVU7Fk@eisner.encompasserve.org>r  d In article <Pbu9a.1$Si.351@news-west.eli.net>, "James T. Sprinkle" <oicmrsnakes@hotmail.com> writes:L > Hi!  I worked with VMS in college several years back on both VAX and AlphaE > architecture.  I am considering getting a hobby copy of VMS and waswI > wondering if it would work on a DecStation?  Also, anybody know where IvK > could look for used VAX/Alpha hardware at a reasonable price?  Thanks fore > any help!t  G    Most "DECstation" are MIPS, not VAX, nor, Alpha.  Please see the FAQt#    (http://www.openvms.compaq.com).h   ------------------------------   Date: 6 Mar 2003 09:48:10 -0800 ) From: P.Young@unsw.EDU.AU (Patrick Young)e Subject: Re: VMSTAR question= Message-ID: <55f85d77.0303060447.377946c5@posting.google.com>   _ "Richard L. Dyson" <rick-dyson@uiowa.edu> wrote in message news:<3E66CD2C.9030305@uiowa.edu>...  > O > If the VMSTAR problem for OpenVMS v7.3-1 has been fixed, would someone pleasei > post here! >  > rick  D Sorry to all folks for not picking up this thread earlier. I've beenE getting too far away from any programming work for quite a while now,tE and C.O.V, (we renovated the top story of our house some time ago andm? are well into the bottom story, as money allows. This plus workpD commitments, tend to leave me too tired - I find myself sleeping far too much on weekends).  D There was that IrDA project for my watch camera I have not looked atC in months .... and OpenVMS tar. I've added quite some stuff such as  GNU tar longlink support.e  B This is all totally untested and like that project un-looked-at in quite some time.  ' The problem you refer to I asked about:c& Google search .... "Broken SYS$SEARCH"  A Someone at Compaq/HP fixed this in the GNV version in the OpenVMS  7.3-1 distribution kit.T  @ Again sorry, I want to get back into this, hopefully winter will& bring on a bit more energy on my part.   ------------------------------  $ Date: Thu, 6 Mar 2003 14:26:57 +05304 From: Kesav Tadimeti <Kesav_Tadimeti@KeaneIndia.com> Subject: Re: [OT]:to SueK Message-ID: <8EA11405E59BD611BA7100104B93C26001009A89@exdel01.del.mgsl.com>    Hi Sue,O Well, L It was given by HP. PLease refer to Pg 26. It is mostly about HP-Oracle. But/ there is a boxed section on HP-UX 11i as well.  K And HP doesn't mention VMS & NSK while they mention HP-UX, Tru64, Linux andr	 Windows.  	 Cheers...c7 +-----------------------------------------------------+s0 	KEANE INDIA LIMITED                            0 	E9 - E12, SDF                                  1 	NEPZ, NOIDA 201 305                             o0 	U.P, INDIA                                     6                                                       0 	e-mail: kesav_tadimeti@keaneindia.com          0 	phone: +91-120-2568210(371)                    5       Men are from MACs, Women are from VMS          i7 +-----------------------------------------------------+n@ Horses are forbidden to eat fire hydrants in Marshalltown, Iowa.   ------------------------------   End of INFO-VAX 2003.128 ************************  H If John can say "No one else will be creating a file at the same time I @ might be," then fine, maybe a file scan is a way to do it.  But G otherwise I say the door is open to getting version number that is NOT tE the same as the file he just opened.  The USEROPEN routine overcomes t this scenario.   Chrisr -----h Chris Olive  System (3424 bytes) started.: >>> 226 Transfer completed.  2720 (8) bytes transferred.  <<< PORT 136,165,113,172,4,2496 >>> 200 Port 4.249 at Host 136.165.113.172 accepted. <<< RETR win_nt_filemon.txtte >>> 150 IMAGE retrieve of /disk$misc/decus/vmslt96a/net96a/win_nt_filemon.txt (3901 bytes) started.s: >>> 226 Transfer completed.  3267 (8) bytes transferred.  <<< PORT 136,165,113,172,4,2506 >>> 200 Port 4.250 at Host 136.165.113.172 accepted. <<< RETR win_95_is_msdos.txtg >>> 150 IMAGE retrieve of /disk$misc/decus/vmslt96a/net96a/win_95_is_msdos.txt (10262 bytes) started. ; >>> 226 Transfer completed.  10218 (8) bytes transferred.t  <<< PORT 136,165,113,172,4,2516 >>> 200 Port 4.251 at Host 136.165.113.172 accepted.2 <<< RETR where_to_get_serial_ports_for_alpha.txtz >>> 150 IMAGE retrieve of /disk$misc/decus/vmslt96a/net96a/where_to_get_serial_ports_for_alpha.txt (1982 bytes) started.: >>> 226 Transfer completed.  1058 (8) bytes transferred.  <<< PORT 136,165,113,172,4,2526 >>> 200 Port 4.252 at Host 136.165.113.172 accepted. <<< RETR what_w95_is.zipd >>> 150 IMAGE retrieve of /disk$misc/decus/vmslt96a/net96a/what_w95_is.zip (105623 bytes) started.< >>> 226 Transfer completed.  105321 (8) bytes transferred.  <<< PORT 136,165,113,172,4,2536 >>> 200 Port 4.253 at Host 136.165.113.172 accepted./ <<< RETR what_http_can_send_automatically.txtaw >>> 150 IMAGE retrieve of /disk$misc/decus/vmslt96a/net96a/what_http_can_send_automatically.txt (1837 bytes) started.2: >>> 226 Transfer completed.  1235 (8) bytes transferred.  <<< PORT 136,165,113,172,4,2546 >>> 200 Port 4.254 at Host 136.165.113.172 accepted. <<< RETR w95filemon.zipsb >>> 150 IMAGE retrieve of /disk$misc/decus/vmslt96a/net96a/w95filemon.zip (36670 bytes) started.; >>> 226 Transfer completed.  36034 (8) bytes transferred..  <<< PORT 136,165,113,172,4,2556 >>> 200 Port 4.255 at Host 136.165.113.172 accepted. <<< RETR w95_memory_use.txtpf >>> 150 IMAGE retrieve of /disk$misc/decus/vmslt96a/ne