1 INFO-VAX	Sat, 02 Dec 2000	Volume 2000 : Issue 672       Contents:3 Re: Charon-VAX Hobbyist Edition - Printout problems " Re: DCL Error Aborts DEC C Install" Re: DEC C V6.2-008 and PGP problem% DEC Side panel needed/TS05 available? ) Re: DEC Side panel needed/TS05 available?  Re: DECNET Proxy Accounts  Re: DECNET Proxy Accounts 4 Ethernet adapter compatibility under VMS 7.2-1/Alpha Re: Linking Option on Alpha. Re: Linking Option on Alpha.$ Re: Mobile code, unsafe at any speed Re: Mozilla M18 i005 Re: Mozilla M18 i005 Re: Mozilla M18 i005 Re: Mozilla M18 i005 (OT to NT)   Re: OpenVMS 7.3 - The right date Re: PAGEFILE resizing/reducion  Re: Possible shared SCSI problem  Re: Possible shared SCSI problem  Re: Possible shared SCSI problem" Re: Postscript to Word conversion?" Re: Tape Drive replacement options RE: Technology of US elections Re: VMS and NT integration?????  Re: VMS and NT integration?????  Re: VMS and NT integration?????  Re: VMS and NT integration?????  Re: VMS vs. Tru64 Unix Re: VMS vs. Tru64 Unix Re: VMS vs. Tru64 Unix Re: VMS vs. Tru64 Unix Re: VMS vs. Tru64 Unix Re: VMS vs. Tru64 Unix Re: VMS vs. Tru64 Unix* Re: Who =?iso-8859-1?Q?=B4s?= the wizard ?* Re: Who =?iso-8859-1?Q?=B4s?= the wizard ? Re: Who s the wizard ?  Re: Who s the wizard ? - Re: Why so much virtual memory for tiny jobs? - Re: Why so much virtual memory for tiny jobs?   F ----------------------------------------------------------------------   Date: 30 Nov 2000 17:23:09 EST1 From: byer@mail.ourservers.net (Robert Alan Byer) < Subject: Re: Charon-VAX Hobbyist Edition - Printout problems1 Message-ID: <oOYF9sLFtgnb@cartman.ourservers.net>    > 8 > That's not very fair - how are we supposed to learn if% > we can't lay hands on the real kit?  > D > If you really upset him he might end up choosing ... dare I say it > ... Solaris? >   3 At least you didn't say the "L" word, being "Linux"   * Ewww, I feel unclean now just typing it :}   --    @  +------------------+--------------------------+---------------+@  | Robert Alan Byer | byer@mail.ourservers.net | ICQ #65926579 |@  +------------------+--------------------------+---------------+@  | Send an E-mail request to obtain a copy of my PGP key.      |@  +-------------------------------------------------------------+@  | "It is by caffeine alone I set my mind in motion.  It is by |@  |  cans of cola the thoughts aquire speed, the hands aquire   |@  |  shakes, the shakes become a warning.  It is by caffeine    |@  |  alone I set my mind in motion."                            |@  +-------------------------------------------------------------+   ------------------------------  # Date: Fri, 01 Dec 2000 19:09:42 GMT 4 From: LESLIE@209-16-45-102.insync.net (Jerry Leslie)+ Subject: Re: DCL Error Aborts DEC C Install ( Message-ID: <WHSV5.276$XY5.10233@insync>  3 Hoff Hoffman (hoffman@xdelta.zko.dec.nospam) wrote:  : H :   I'm aware of a previous report of a similar problem with V6.2A (thatI :   exists up through FT2 T6.4 compiler), and a potential workaround for  E :   that problem is to answer "yes" to the license prompt and then to C :   ignore any IVP-related failure messages that might later arise.  :    The workaround worked.  / --Jerry Leslie   leslie@209-16-45-97.insync.net ;                  leslie@209-16-45-102.insync.net is invalid 2                  (my opinions are strictly my own)   ------------------------------  % Date: Sat, 02 Dec 2000 04:12:31 +0100 2 From: martin@radiogaga.harz.de (Martin Vorlaender)+ Subject: Re: DEC C V6.2-008 and PGP problem ; Message-ID: <3a28689f.524144494f47414741@radiogaga.harz.de>   + Chris De Young (chd@vms.arizona.edu) wrote: I : So, I'm not much of a C guru by any stretch of the imagination, but I'm J : trying to compile PGP 2.6.3i under VMS AXP V7.1-1H2, and the problem I'm : having is that PGP.C says: :  : #include <ctype.h> : ' : and SYS$SYSROOT:[SYSLIB]CTYPE.H says:  :  : #include "common_header.h" : ; : and no "common_header.h" exists anywhere that I can find.   D You should be aware that the DEC C compiler doesn't use header filesF in SYS$LIBRARY: (BTW: I only remember them being in that location in aD VAX C environment). Instead, it's using SYS$LIBRARY:DECC$RTL.TLB. In there, sure enough:   0 $ library/list/names sys$library:decc$rtldef.tlb> Directory of TEXT library SYS$COMMON:[SYSLIB]DECC$RTLDEF.TLB;1 ...  COBDEF COMMON_HEADER		<--- 
 CONV$ROUTINES  ...   J At installation time, you have the option to install reference copies intoJ SYS$COMMON:[DECC$LIB.REFERENCE...]. To repeat it: these are only reference& copies, the compiler doesn't use them.  ) :  Am I missing a piece of my C compiler?   @ I don't know - you didn't tell us which C compiler you're using.   cu,    Martin --J One OS to rule them all       | Martin Vorlaender  |  VMS & WNT programmer7 One OS to find them           | work: mv@pdv-systeme.de N One OS to bring them all      |       http://www.pdv-systeme.de/users/martinv/> And in the Darkness bind them.| home: martin@radiogaga.harz.de   ------------------------------  # Date: Sat, 02 Dec 2000 02:26:17 GMT / From: StevenU@POBoxes.com (Steven P. Underwood) . Subject: DEC Side panel needed/TS05 available?2 Message-ID: <3a2859eb.171642596@news.telocity.com>  
 Hello all:  E I have rescued(?) an empty equipment rack at work for reuse but it is B missing a side panel. I would like to find one in the southern NewC England area, if possible, as this venture is not worth the cost of A shipping such a large part.  Also, my wife does not want any more C "junk" in the house, but if it is complete, I can easily argue that 
 point. ;7)  D  The model number on the unit is H9542FB and is a 40" high rack thatD originally housed a TS05 tape drive (hinged top).  The side panel isB 30" x 39.5". If anyone in the area has one of these hanging aroundE (possibly after fusing 2 units, which is why this one is missing it's > side) and would be willing to part with it, please contact me.  D Also, if anyone has a need for the TS05 tape drive/controller/cablesD that has not been used in many moons (>5 years), I could discuss theD availability of this unit with my boss. I can not make any promises,E however he did ask me to inquire (unsuccessfully) with several of our  vendors.  	 Thank you      Steven P. Underwood,DNRC Whitinsville,MA  StevenU@POBoxes.com    ------------------------------  % Date: Sat, 02 Dec 2000 00:16:12 -0500 - From: JF Mezei <jfmezei.spamnot@videotron.ca> 2 Subject: Re: DEC Side panel needed/TS05 available?, Message-ID: <3A288595.18E04B9B@videotron.ca>   "Steven P. Underwood" wrote:G > I have rescued(?) an empty equipment rack at work for reuse but it is  > missing a side panel.   L Are these the fat panels with the airvents on one side that were used on theK Microvax II large cabinets as well as the RA82 cabs as well ? If so, do you  neex the left or right one ?   ------------------------------   Date: 1 Dec 2000 18:51:28 GMT 2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)" Subject: Re: DECNET Proxy Accounts6 Message-ID: <908rvg$ood$2@mailint03.im.hou.compaq.com>  Z In article <3A27BD2B.30832AB0@fmr.com>, "David J. Pulling" <david.pulling@fmr.com> writes:I :When converting to Decnet Phase v or OSI the old proxy file netproxy.dat < :still contains old proxy accounts. How do you remove them ?  I   NET$PROXY.DAT and NETPROXY.DAT exist on all OpenVMS VAX V6.1 and later  G   and on all OpenVMS Alpha V6.2 and later systems.  The presence of the G   NET$PROXY.DAT file is not specific to DECnet Phase IV or DECnet-Plus.   ,   Specific AUTHORIZE commands involved here?  6   Specific proxy entries that are causing the problem?  I   If this is actually OpenVMS Alpha V7.2-1, I'd get the customer support  K   center involved.  (I've seen this a few times on other OpenVMS releases,  J   but it is rare on more recent releases.  That said, I would like to see K   a REMOVE/PROXY/OLD added to AUTHORIZE.)  This may well wind up involving  8   an ECO to AUTHORIZE or the underlying system services.  G   I'll ask that the (duplicate) discussion of this over in the Ask The     Wizard area be removed.   N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Fri, 01 Dec 2000 20:14:18 -0600 7 From: "David J. Dachtera" <djesys.nospam@earthlink.net> " Subject: Re: DECNET Proxy Accounts- Message-ID: <3A285AFA.4B181C01@earthlink.net>    "David J. Pulling" wrote:  > + > When converting to Decnet Phase v or OSI     Why?  ! > the old proxy file netproxy.dat = > still contains old proxy accounts. How do you remove them ?    Perhaps a little DCL...    --   David J. Dachtera  dba DJE Systems  http://www.djesys.com/  : Unofficial Affordable OpenVMS Home Page and Message Board: http://www.djesys.com/vms/soho/   F This *IS* an OpenVMS-related newsgroup. So, a certain bias in postings is to be expected.  @ Feel free to exercise your rights of free speech and expression.  F However, attacks against individual posters, or groups of posters, are strongly discouraged.    ------------------------------  % Date: Fri, 01 Dec 2000 23:43:11 -0500  From: alanb@cloud9.net= Subject: Ethernet adapter compatibility under VMS 7.2-1/Alpha * Message-ID: <3A287DDE.6A07CE3C@cloud9.net>  B Has anyone tried any 3COM card on an Alpha? The SPD only lists non third-party ones.    ------------------------------  # Date: Fri, 01 Dec 2000 21:21:25 GMT , From: "J. Scott Greig" <jsgreig@geminaq.com>% Subject: Re: Linking Option on Alpha. 3 Message-ID: <pDUV5.8118$7.265735@quark.idirect.com>   4 "Michael" <maustin@nc.prestige.net> wrote in message) news:3A27D4AB.CFA47893@nc.prestige.net... I > I am creating a shared image with several functions.  What is the Alpha $ > equivalent to the VAX link option: >  > universal=<routine1> > universal=<routine2>   You will want the following:  & symbol_vector= (<routine1>=procedure,-3                              <routine2>=procedure,-                                .                                . 4                               <data_element1>=data,-4                                <data_element2>=data)  F I have found that data elements (globals) if any, are best if at least quadword aligned.    Hope this helps  Scott    > G > I do not have access to the doc set and it has been at least a couple 5 > years and several contracts since I had to do this.  > 	 > Thanks,  > Michael Austin > DBA Consultant.  >    ------------------------------   Date: 1 Dec 2000 22:12:01 GMT 2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)% Subject: Re: Linking Option on Alpha. 6 Message-ID: <9097nh$qe6$1@mailint03.im.hou.compaq.com>  Y In article <3A27D4AB.CFA47893@nc.prestige.net>, Michael <maustin@nc.prestige.net> writes: H :I am creating a shared image with several functions.  What is the Alpha :equivalent to the VAX...   D   Please check the OpenVMS FAQ for a pointer to the shareable image    cookbook.   ' :I do not have access to the doc set...   I   I would expect that you could gain access to the on-line (HTML-format)  K   documentation for OpenVMS.  Please check the OpenVMS FAQ for the on-line  0   version URL and for other related information.  "   The OpenVMS FAQ is available at:"     http://www.openvms.compaq.com/  (   OpenVMS documentation is available at:'     http://www.openvms.compaq.com:8000/   N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Fri, 01 Dec 2000 16:46:57 -0500 , From: "Glenn C. Everhart" <Everhart@GCE.com>- Subject: Re: Mobile code, unsafe at any speed & Message-ID: <3A27D601.5912BEA@GCE.com>  5 A question in my mind is whether sandboxing is really @ equivalent to being treated as a different user. If my sandboxed= program creates another program, I'd expect it normally to be = marked as having been created by my identity (instead of some > "foreign alien enemy" identity, perhaps.) If a program runs as> a different user with different permissions, things it creates: tend to be marked as that (alien) user's, and thus keep on< being treated as untrusted if executed later, unless someone
 changes this.   < Sandboxing is to my mind a kludgey way of faking, partially,A separate identities. It gets used IMO either because people don't > have the ability to treat multiple identities, or because they? don't want the overhead of setting up access for alien versions = of all users. I also have to wonder whether installing images = like browsers with resource identifiers that can give partial ? identity tags to stuff they create might be a way to accomplish - the sandboxing kind of access in vanilla VMS.      Nigel Arnot wrote: >  > > J > > "Larry Kilgallen" <Kilgallen@eisner.decus.org.nospam> wrote in message) > > news:GOSi3uE3E2RV@eisner.decus.org... C > > > > There are several ways of getting code down to client PC's:  > > > G > > > But the very idea of running code provided by an untrusted source  > > > is flawed from the start.  > > G > > I don't think the distinction is as clearcut as you make it between = > > markup languages and code. To take an example: PS viewers = > > have security problems. Even VT terminals have well-known  > > security problems. > >  > M > It's not a matter of whether it's a "markup" or not, it a matter of whether K > the untrusted code runs in a perfectly virtualized sandbox or not. Vmware Q > (www.vmware.com) is a rather good virtualisation of a complete PC, that is used I > to develop things like kernels, filesystems, and antivirus software, so K > that mistakes reversibly splat a virtual machine rather than irreversibly K > splatting the developer's real one. IBM's VM is another example, recently M > reported as happily running 42000 virtual Linuxes in a single IBM dinosaur.  > F > If a virtualisation doesn't have any serious design flaws, it's mostO > unlikely that an accidental piece of bad code can do any harm. More dangerous G > is code written by a cracker with the intent and ability to exploit a M > bug in the virtual machine: whether that's a serious risk boils down to how B > much you trust the VM and the O/S underwhich it runs. There's noI > intrinsic reason why running untrusted code in a sandbox is any riskier K > than having a real O/S (VMS!) connected to the internet. Both are safe in + > theory, but may be crackable in practice.  >  >         Yours, >                 Nigel Arnot * >                 NRA@MAXWELL.PH.KCL.AC.UK > G >                 "In the beginning there was nothing, which exploded."    ------------------------------  # Date: Fri, 01 Dec 2000 21:47:37 GMT ' From: Colin Blake <colin@theblakes.com>  Subject: Re: Mozilla M18 i005 , Message-ID: <3A281BDA.7C514A9@theblakes.com>   Brian Tillman wrote:  9 > >Java in OpenVMS Mozilla isn't supported at the moment.  > N > Then why is the Q insisting that it's for OpenVMS Alpha only?  With no Java,2 > it SHOULD run on a VAX as well.  Why doesn't it?  - VAX doesn't have upcalls support, that's why.    ------------------------------  % Date: Fri, 01 Dec 2000 19:12:28 -0600 7 From: "David J. Dachtera" <djesys.nospam@earthlink.net>  Subject: Re: Mozilla M18 i005 - Message-ID: <3A284C7C.DB135D67@earthlink.net>    Colin Blake wrote: >  > Brian Tillman wrote: > ; > > >Java in OpenVMS Mozilla isn't supported at the moment.  > > P > > Then why is the Q insisting that it's for OpenVMS Alpha only?  With no Java,4 > > it SHOULD run on a VAX as well.  Why doesn't it? > / > VAX doesn't have upcalls support, that's why.   $ O.k. I'll bite - what are "upcalls"?   --   David J. Dachtera  dba DJE Systems  http://www.djesys.com/  : Unofficial Affordable OpenVMS Home Page and Message Board: http://www.djesys.com/vms/soho/   F This *IS* an OpenVMS-related newsgroup. So, a certain bias in postings is to be expected.  @ Feel free to exercise your rights of free speech and expression.  F However, attacks against individual posters, or groups of posters, are strongly discouraged.    ------------------------------   Date: 1 Dec 2000 21:12:29 -0500 9 From: Kilgallen@eisner.decus.org.nospam (Larry Kilgallen)  Subject: Re: Mozilla M18 i0052+ Message-ID: <DeJ6NVxiz0eH@eisner.decus.org>   g In article <3A284C7C.DB135D67@earthlink.net>, "David J. Dachtera" <djesys.nospam@earthlink.net> writes:8 > Colin Blake wrote: >> p >> Brian Tillman wrote:e >> e< >> > >Java in OpenVMS Mozilla isn't supported at the moment. >> >Q >> > Then why is the Q insisting that it's for OpenVMS Alpha only?  With no Java, 5 >> > it SHOULD run on a VAX as well.  Why doesn't it?h >> e0 >> VAX doesn't have upcalls support, that's why. > & > O.k. I'll bite - what are "upcalls"?  H Upcalls are a mechanism for VMS to tell DECthreads that a system serviceE call will "take a while" and switching threads would allow fuller CPUo utilization.   	HELP LINK/THREADS_ENABLEa   ------------------------------    Date: 01 Dec 2000 14:53:46 -0600- From: Graham Allan <allan@mnhep1.hep.umn.edu>o( Subject: Re: Mozilla M18 i005 (OT to NT)0 Message-ID: <w53itp3lv6t.fsf@lanark.spa.umn.edu>  1 Nigel Arnot <sysmgr@maxwell.ph.kcl.ac.uk> writes:M  M > Actually, not as bad as you think. It's only a single file, nsreg.dat, that6O > requires write access. Write-protect the directory, then unprotect that file.e > H > The thing I find more infuriating is that there's no way to lock usersG > out of the profile manager. (I create a profile "Network User" storedcK > on the user's home share, but every so often some idiot gets into profilesK > manager and deletes of mangles that profile rather than simply using edit K > options in netscape to set up his personal profile whatever way he wants)   I Yes, same here, or very occasionally someone logs into a workstation at a-H time their home share isn't available, so Netscape happily creates a new# profile back under Program Files...e   Graham -- sI -------------------------------------------------------------------------.: Graham Allan - I.T. Manager - gta@umn.edu - (612) 624-50409 School of Physics and Astronomy - University of MinnesotawI --------------------------------------------------------------------------   ------------------------------  % Date: Fri, 01 Dec 2000 20:16:37 -060057 From: "David J. Dachtera" <djesys.nospam@earthlink.net>+) Subject: Re: OpenVMS 7.3 - The right date - Message-ID: <3A285B85.A6D5D96E@earthlink.net>-  * fabio_compaq@ep-bc.petrobras.com.br wrote: > H > Do you know what date the OVMS 7.3 will be release ? Jan, Feb or Mar ?2 > The online documentation just say First Quarter.  E First you're anxious to get rid 'em, now you're antsy to upgrade 'em?   + Do you want it now or do you want it right?    --   David J. Dachtera- dba DJE Systems- http://www.djesys.com/  : Unofficial Affordable OpenVMS Home Page and Message Board: http://www.djesys.com/vms/soho/r  F This *IS* an OpenVMS-related newsgroup. So, a certain bias in postings is to be expected.  @ Feel free to exercise your rights of free speech and expression.  F However, attacks against individual posters, or groups of posters, are strongly discouraged.h   ------------------------------  # Date: Fri, 01 Dec 2000 21:13:01 GMTo* From: Doran Werling <rwscsinc@my-deja.com>' Subject: Re: PAGEFILE resizing/reducion ) Message-ID: <90948l$sap$1@nnrp1.deja.com>e  L In article <OFB4F50B18.9FBAADD9-ON032569A8.0064CAE1@ep-bc.petrobras.com.br>,,   fabio_compaq@ep-bc.petrobras.com.br wrote: >sD > If I have two or more PAGEFILEs spreaded across my disk, and theseB > files are not really being used as seen below, may I consider toA > reduce or eliminate one of them ? Or it's technically better tot > maintain them divided ?: >CH > PAGEFILE1_SIZE information (for DISCO01:[SYS0.SYSEXE]PAGEFILE1.SYS;1): >         Feedback information.g8 >            Old value was 1679400, New value is 20985004 >            Maximum observed usage (blocks): 821792C >            PAGEFILE1_SIZE will be modified to hold 2098500 blocksm > H > PAGEFILE2_SIZE information (for DISCO02:[SYS0.SYSEXE]PAGEFILE2.SYS;1): >         Feedback information.i8 >            Old value was 1679400, New value is 20985004 >            Maximum observed usage (blocks): 703504C >            PAGEFILE2_SIZE will be modified to hold 2098500 blocksB > H > PAGEFILE3_SIZE information (for DISCO03:[SYS0.SYSEXE]PAGEFILE3.SYS;1): >         Feedback information.R8 >            Old value was 1679400, New value is 20985004 >            Maximum observed usage (blocks): 676528C >            PAGEFILE3_SIZE will be modified to hold 2098500 blocks- >  > Fabio Cardoso. >  >t  F Take a look at what autogen is telling you.  When the utilization of aD paging file exceeds 50%, internal page file fragmentation becomes anD issue. This is why autogen wants to increase the size of your paging? files. Your page files are on that threshold right now. I would G personally leave them unchanged for now.  If your system load is fairly 1 static.  Otherwise consider increasing the sizes.    Regards,
 Doran Werling  RW/SCS Inc.s    & Sent via Deja.com http://www.deja.com/ Before you buy.p   ------------------------------  % Date: Fri, 01 Dec 2000 14:40:52 -0500r& From: Donald G Plugge <plugge@usa.net>) Subject: Re: Possible shared SCSI problem1' Message-ID: <3A27FEC2.A23ABB2B@usa.net>    Uwe Zessin wrote:c  ) > In article <3A27CE60.32FD0291@usa.net>,s >   plugge@usa.net wrote:@E > > I have two DS10's sharing a SCSI bus on which the system drive ist	 > snip...r6 > > Problem: System Delay possibly due to SCSI adapter >t > Excellent guess ! See below: >l > > Details: > >l > > 2 - DS10 OpenVMS 7.2-1 > > 2 - KZPBA-CA PCI-UltraSCSI >a
 > Too bad:J > http://www.compaq.com/alphaserver/options/asds10/asds10___-kzpba-ca.html > says:a
 > "* For VMS:t5 >           * OpenVMS SCSI Clustering not supported."5  E Sorry, my mistake.  My internal drives are on a KZPBA-CA, however, mynI differential SCSI box is on a KZPBA-CB (the docs say -CB, but the card is  marked as -CY).u   >p >iH > You will need 2 KZPBA-CB adapters, two differential terminators and... >o2 > > 1 - UltraSCSI RAID SBB Shelf (DS-BA356-Series) >lF > ... a differential I/O module. A DS-BA35X-DA, if I recall correctly.E > Please check the documentation that (hopefully) came with your box.a  	 Got that!s   >  >hF > Do not forget to assign different SCSI IDs to both KZPBA-CB (7 + 6).E > If you need a drive in slot 6, then you can turn on high-addressingv@ > for the box - disks will start with SCSI ID 8 and go up to 14.2 > Again, this is in the manual for the I/O module.  
 Already done!    >m >F > > Allocation Classes:i > >o( > > allocation class = 5 (on both nodes) > > - > > port allocation = 0  (for PKA non-shared)o) > > port allocation = 10 (for PKB shared)  > [...]s >d< > Next, I increment the allocation class for each bus. E.g.: > Host 1, allocation class = 10  >   PKA0, " = 11 (internal)n >   PKB0, " = 12 (tape)u >   PKC0, " = 1  (shared bus)a > ...a > Host 2, allocation class = 20  >   PKA0, " = 21 (internal)  >   PKB0, " = 1  (shared bus)- >   PKC0 - does not exist   : So you are saying that I should change my MODPARAMS.DAT to8 for ALLOCLASS=5 on both nodes to ALLOCLASS=10 on one andE ALLOCLASS=20 on another.  The docs say to use the same for each node.3  . Then in my SYS$DEVICE.DAT I should change from   [Port NODE1$PKA] Allocation Class = 0   [Port NODE1$PKB] Allocation Class = 10r   [Port NODE2$PKA] Allocation Class = 0   [Port NODE2$PKB] Allocation Class = 10.   to something like...   [Port NODE1$PKA] Allocation Class = 11n   [Port NODE1$PKB] Allocation Class = 1   [Port NODE2$PKA] Allocation Class = 21    [Port NODE2$PKB] Allocation Class = 1  H This seems at odds with the PAC writeup that Digital sent me, but it may
 just work.   >a > Good luck! >  > -- > Uwe Zessin >s( > Sent via Deja.com http://www.deja.com/ > Before you buy.    ------------------------------  # Date: Fri, 01 Dec 2000 21:05:20 GMTe% From: Uwe Zessin <zessin@my-deja.com>e) Subject: Re: Possible shared SCSI problem ) Message-ID: <9093qa$rsb$1@nnrp1.deja.com>i  = In article <3A27FEC2.A23ABB2B@usa.net>, plugge@usa.net wrote:r >n > Uwe Zessin wrote:  >'A > > In article <3A27CE60.32FD0291@usa.net>, plugge@usa.net wrote: G > > > I have two DS10's sharing a SCSI bus on which the system drive isl > > snip...-8 > > > Problem: System Delay possibly due to SCSI adapter > >i  > > Excellent guess ! See below: > >  > > > Details: > > >- > > > 2 - DS10 OpenVMS 7.2-1  > > > 2 - KZPBA-CA PCI-UltraSCSI > >h > > Too bad:E > > http://www.compaq.com/alphaserver/options/asds10/asds10___-kzpba-  ca.html.	 > > says:r > > "* For VMS:D7 > >           * OpenVMS SCSI Clustering not supported."  >sD > Sorry, my mistake.  My internal drives are on a KZPBA-CA, however,B > my differential SCSI box is on a KZPBA-CB (the docs say -CB, but > the card is marked as -CY).D   -CY is just the logic moduleD -CB is the entire package (logic module, documentation, floppy disc)  7 (And -CX is the logic module from the KZPBA-CA package)n  C > > You will need 2 KZPBA-CB adapters, two differential terminatorsp
 > > and... > >e4 > > > 1 - UltraSCSI RAID SBB Shelf (DS-BA356-Series) > > H > > ... a differential I/O module. A DS-BA35X-DA, if I recall correctly.G > > Please check the documentation that (hopefully) came with your box.e >0 > Got that!0 > H > > Do not forget to assign different SCSI IDs to both KZPBA-CB (7 + 6).G > > If you need a drive in slot 6, then you can turn on high-addressingtB > > for the box - disks will start with SCSI ID 8 and go up to 14.4 > > Again, this is in the manual for the I/O module. >  > Already done!  >  > > > Allocation Classes:s > > >e* > > > allocation class = 5 (on both nodes) > > > / > > > port allocation = 0  (for PKA non-shared)m+ > > > port allocation = 10 (for PKB shared)t	 > > [...]e > >s> > > Next, I increment the allocation class for each bus. E.g.:! > > Host 1, allocation class = 10t > >   PKA0, " = 11 (internal)  > >   PKB0, " = 12 (tape)  > >   PKC0, " = 1  (shared bus)= > > ...>! > > Host 2, allocation class = 20  > >   PKA0, " = 21 (internal)e > >   PKB0, " = 1  (shared bus)  > >   PKC0 - does not exist  > < > So you are saying that I should change my MODPARAMS.DAT to: > for ALLOCLASS=5 on both nodes to ALLOCLASS=10 on one andG > ALLOCLASS=20 on another.  The docs say to use the same for each node.e  B No, I'm not saying you should do it. My intention was to tell whatA I have done in the past to avoid naming conflicts with IDE CD-ROMe? drives, because (last time I checked) it was/is not possible tot use PACs on them.   0 > Then in my SYS$DEVICE.DAT I should change from'                               could ;-)-   >  > [Port NODE1$PKA] > Allocation Class = 0 [...]l > to something like... >f > [Port NODE1$PKA] > Allocation Class = 11  [...]-  = Yes. That looks similar to what I have done with some systems7> in the past. I like running all disks with allocation classes.= If you, one day for some reason, decide to shadow with a disks2 on a local bus you need an allocation class on it.  B You are free to 'adopt' my idea, but I didn't intend to force you.  #                                 ---o  " I have some questions - you wrote: > For instance, I can issue a: >- > $ show dev d  9 Is this really from the DCL level or from console (>>>) ?F? Does the error count increase on disks at the shared bus and/ort# do you get any mount-verifications?e   --
 Uwe Zessin    & Sent via Deja.com http://www.deja.com/ Before you buy.e   ------------------------------    Date: 02 Dec 2000 12:24:21 +0800/ From: Paul Repacholi <prep@k9.prep.synonet.com>-) Subject: Re: Possible shared SCSI problemM0 Message-ID: <87k89jpi16.fsf@k9.prep.synonet.com>  ( Donald G Plugge <plugge@usa.net> writes:   <stuff>:  7 try rebooting with the controller ID swapped over. Prob 6 means setting 6 to 7 and visa versa. If this cures it,7 you have the SCSI arbitration problem, explained in thes; cluster manual. Drop your controller ID below the critical/s  busy drives if this is the case.  9 The problem is, SCSI arbitration is highest ID wins. Withb< big queues in the drive, and lots of buffer a controller can< hog the bus and other units can't make progress. Putting the: controller to a low ID gives the drives priority. You lose8 a bit on the peak performance of one system, but gain if' multiple hosts are saturating the SCSI.i   SCSI... bleh  ; I do hope the great gods of decpaq don't think SAN and SCSIk. can replace a well designed IO arch like MSCP.   ~Paul    ------------------------------  % Date: Fri, 01 Dec 2000 20:09:07 -0600p7 From: "David J. Dachtera" <djesys.nospam@earthlink.net>f+ Subject: Re: Postscript to Word conversion?l- Message-ID: <3A2859C3.79671050@earthlink.net>g   David Mathog wrote:  > I > In article <908gr8$aku$1@nnrp1.deja.com>, byatesiii@my-deja.com writes:r > >mE > >Hmm, Ghostscript for VMS? I searched long and hard for postscript,VB > >found alot of unix/linux stuff, but no VMS. Will try again! Got > >pointers, anyone? > 5 >   http://www.cs.wisc.edu/~ghost/aladdin/get601.htmls > = > It needed a few minor tweaks to build.  Contact me by emaili > if you have a problem.  D Can you post the VMS binaries? ...an object library or two at least?  ) I have server space, if that's a problem.'  hH > Why didn't you say so in the first place?  Ghostscript will definitelyD > convert Postscript to PDF - I use it for that all the time.   Try: > Q >  $ gs "-sDEVICE=pdfwrite" "-dNOPAUSE" "-sOutputFile=yourchoice.pdf" yourfile.pst  H I don't have a link to xv right now, but doesn't that come with a ps2pdf utility?   -- n David J. Dachterar dba DJE Systemso http://www.djesys.com/  : Unofficial Affordable OpenVMS Home Page and Message Board: http://www.djesys.com/vms/soho/   F This *IS* an OpenVMS-related newsgroup. So, a certain bias in postings is to be expected.  @ Feel free to exercise your rights of free speech and expression.  F However, attacks against individual posters, or groups of posters, are strongly discouraged.u   ------------------------------  % Date: Fri, 01 Dec 2000 06:15:50 -0500s) From: yyyc186.illegaltospam_@flashcom.nett+ Subject: Re: Tape Drive replacement optionsy9 Message-ID: <3a2788a5$1$lllp186$mr2ice@news.flashcom.com>   J In <4%iV5.2579$Ei1.159309@bgtnsc05-news.ops.worldnet.att.net>, on 11/30/00  ?    at 02:32 AM, "Robert Meyer" <meyerra@worldnet.att.net> said:   G I have a Sony SDT-7000 tape drive in my PC-164 running the Hobbiest VMSu" license.  Seems to work just fine.   Roland  F >Our Alpha server is on it's 3rd tape drive from Compaq. The tapes areI >new, and Compaq sends a Digital service person.   I have an external 4mm E >HP DAT DDS II tape drive.  Does the Alpha recognize an HP, and if it < >does, can I get drivers to connect, or am I limited to onlyI >Compaq/Digital.  I also have an older DLT III.  If I connect this onto a3H >SCSI Channel, will the system recognize, if not are drivers available? 1 >Thank you in advance for your time and patience.>     --  ; -----------------------------------------------------------aD yyyc186@flashcom.net              To Respond delete ".illegaltospam"6                             MR/2 Internet Cruiser 2.2a8                             For a Microsoft free univers; -----------------------------------------------------------r   ------------------------------  % Date: Fri, 01 Dec 2000 14:17:48 -0500r4 From: "Bochnik, William J" <BochnikWJ@bernstein.com>' Subject: RE: Technology of US elections J Message-ID: <2B37459189B0D211BE710000F8EF9D8508908927@nts0147.beehive.com>  J This message is in MIME format. Since your mail reader does not understand< this format, some or all of this message may not be legible.  ' ------_=_NextPart_001_01C05BCB.62DB5E2A  Content-Type: text/plain;s 	charset="iso-8859-1"   E actually they might save us a lot more time/money/embarassment if the / truck(s) caught fire and the ballots were lost.f   -----Original Message-----4 From: JF Mezei [mailto:jfmezei.spamnot@videotron.ca] Sent: December 01, 2000 9:23 AMo To: Info-VAX@mvb.saic.como' Subject: Re: Technology of US electionso    J Watching the OJ Simpson re-creation yesterday with that Ryder rent-a-truckH driving on the roads, with helicopters constantly filming it, I couldn't helpI but wish a serious accident where the punched cards would be expelled ande flynL into the wind, spreading all over the area. (Of course, broadcasted live :-)  I Think of the hype the media would have, and think about all the $$$ signsl thatH the lawyers would see in trying to deal with the situation where ballots fromH one county would be completely lost/unusable and thus the state would be' unable to fullfil its own requirements.> ...   ' ------_=_NextPart_001_01C05BCB.62DB5E2AS Content-Type: text/html; 	charset="iso-8859-1"o+ Content-Transfer-Encoding: quoted-printableV  1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">. <HTML> <HEAD>9 <META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =l charset=3Diso-8859-1">@ <META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version =
 5.5.2651.65">o- <TITLE>RE: Technology of US elections</TITLE>  </HEAD>- <BODY>  : <P><FONT SIZE=3D2>actually they might save us a lot more =E time/money/embarassment if the truck(s) caught fire and the ballots =o were lost.</FONT>w </P>  3 <P><FONT SIZE=3D2>-----Original Message-----</FONT>w' <BR><FONT SIZE=3D2>From: JF Mezei [<A = I HREF=3D"mailto:jfmezei.spamnot@videotron.ca">mailto:jfmezei.spamnot@vide=r otron.ca</A>]</FONT>9 <BR><FONT SIZE=3D2>Sent: December 01, 2000 9:23 AM</FONT>h3 <BR><FONT SIZE=3D2>To: Info-VAX@mvb.saic.com</FONT>yA <BR><FONT SIZE=3D2>Subject: Re: Technology of US elections</FONT>N </P> <BR>  F <P><FONT SIZE=3D2>Watching the OJ Simpson re-creation yesterday with = that Ryder rent-a-truck</FONT>F <BR><FONT SIZE=3D2>driving on the roads, with helicopters constantly =" filming it, I couldn't help</FONT>H <BR><FONT SIZE=3D2>but wish a serious accident where the punched cards =  would be expelled and fly</FONT>D <BR><FONT SIZE=3D2>into the wind, spreading all over the area. (Of =# course, broadcasted live :-)</FONT>. </P>  E <P><FONT SIZE=3D2>Think of the hype the media would have, and think =-# about all the $$$ signs that</FONT>aE <BR><FONT SIZE=3D2>the lawyers would see in trying to deal with the = # situation where ballots from</FONT>-E <BR><FONT SIZE=3D2>one county would be completely lost/unusable and =- thus the state would be</FONT>A <BR><FONT SIZE=3D2>unable to fullfil its own requirements.</FONT>y <BR><FONT SIZE=3D2>...</FONT>c </P>   </BODY>3 </HTML>g) ------_=_NextPart_001_01C05BCB.62DB5E2A--r   ------------------------------  % Date: Fri, 01 Dec 2000 19:43:42 -0600o7 From: "David J. Dachtera" <djesys.nospam@earthlink.net>r( Subject: Re: VMS and NT integration?????- Message-ID: <3A2853CE.AB427FED@earthlink.net>    Jordan Henderson wrote:g1 > [snip]  Putting aside your obvious distaste forrD > deprecated technologies like LAT and Microvaxes, I can think of noC > better way to access an OpenVMS system console remotely except by D > reverse-LAT to a Terminal Server that has a port connected to that > system's console.i  6 I s'pose reverse-TELNET is out of the question then...  G Did ANYone ever figure out how to route LAT (or any other protocol thatd" has no routing layer of its own)?    --   David J. Dachterad dba DJE Systems- http://www.djesys.com/  : Unofficial Affordable OpenVMS Home Page and Message Board: http://www.djesys.com/vms/soho/G  F This *IS* an OpenVMS-related newsgroup. So, a certain bias in postings is to be expected.  @ Feel free to exercise your rights of free speech and expression.  F However, attacks against individual posters, or groups of posters, are strongly discouraged.a   ------------------------------  % Date: Fri, 01 Dec 2000 19:53:57 -0600o7 From: "David J. Dachtera" <djesys.nospam@earthlink.net> ( Subject: Re: VMS and NT integration?????- Message-ID: <3A285635.CAA1EEC2@earthlink.net>S  * fabio_compaq@ep-bc.petrobras.com.br wrote: > A > My problem with the two VAX 4000 , one Microvax 3100,  and  one0
 >  RX400 are:M > D > 1 - They are ocupping space in the data center; We are putting the* >  new servers on the racks (Compaq racks)  E VAX 4000s and MicroVAX 3100s can be put into racks. I've done it. Not  sure exactly what an RX400 is.   G > 2 - They are just Pathworks file server for a group of users. All theoA > company is using WNT file serving, but these users dont want to  >  change.... (???)5  A Both of 'em could be changed to Linux + Samba with a little work.E  G9 > 3 - What means the operators must backup them everyday.   1 So, they're not backing up the WNT servers ???!!!    8 > 4 - When my manager agrees to stop the machines nobodyD > except the file server users will notice, and they can just re-map >  the \\server\share.   What's stopping them?t  iD >  Today, Hoffman put me in my right place ! Forgive me people ! :-)J > I will not make comments about loved legacy hardware and software !  :-)  # Well, this *IS* comp.os.*VMS*, no? u   -- f David J. Dachterar dba DJE Systems  http://www.djesys.com/  : Unofficial Affordable OpenVMS Home Page and Message Board: http://www.djesys.com/vms/soho/i  F This *IS* an OpenVMS-related newsgroup. So, a certain bias in postings is to be expected.  @ Feel free to exercise your rights of free speech and expression.  F However, attacks against individual posters, or groups of posters, are strongly discouraged.h   ------------------------------   Date: 1 Dec 2000 21:13:42 -0500e9 From: Kilgallen@eisner.decus.org.nospam (Larry Kilgallen) ( Subject: Re: VMS and NT integration?????+ Message-ID: <qw3pY2gdhcB+@eisner.decus.org>m  g In article <3A2853CE.AB427FED@earthlink.net>, "David J. Dachtera" <djesys.nospam@earthlink.net> writes:   I > Did ANYone ever figure out how to route LAT (or any other protocol thatP$ > has no routing layer of its own)?   C As in, has anyone figured out how to operate a protocol outside itsa' specified environment so it will fail ?M   ------------------------------   Date: 1 Dec 2000 22:32:02 -0500 / From: jordan@lisa.gemair.com (Jordan Henderson)i( Subject: Re: VMS and NT integration?????) Message-ID: <909qfi$bi$1@lisa.gemair.com>   - In article <3A2853CE.AB427FED@earthlink.net>,d6 David J. Dachtera <djesys.nospam@earthlink.net> wrote: >Jordan Henderson wrote:2 >> [snip]  Putting aside your obvious distaste forE >> deprecated technologies like LAT and Microvaxes, I can think of nopD >> better way to access an OpenVMS system console remotely except byE >> reverse-LAT to a Terminal Server that has a port connected to that  >> system's console. > 7 >I s'pose reverse-TELNET is out of the question then...s >t  C Not at all.  While I could think of no better way than reverse-LAT,eC reverse-TELNET could be a close second.  I might prefer LAT because ) of it's efficiency and it's simplicity.  )  H >Did ANYone ever figure out how to route LAT (or any other protocol that# >has no routing layer of its own)?   >e  D Depends on your needs, of course, but I might argue that the lack ofG routing makes LAT superior for just this application.  With LAT there'suB less danger that a black hat from outside your network could gain @ access to your system console.  Of course, this could also be a B severe limitation in some configurations where routing is required= between where you desire to perform the remote administrationCC and the Terminal Server.  Of course, there's always ssh to a system @ where reverse-LAT could be accessed near enough to your Terminal* Server to provide the best of both worlds.   >--  >David J. Dachtera >dba DJE Systems >http://www.djesys.com/o >i; >Unofficial Affordable OpenVMS Home Page and Message Board:o  >http://www.djesys.com/vms/soho/ >]G >This *IS* an OpenVMS-related newsgroup. So, a certain bias in postingsr >is to be expected.i >CA >Feel free to exercise your rights of free speech and expression.a >dG >However, attacks against individual posters, or groups of posters, area >strongly discouraged.   -Jordan Henderson  jordan@greenapple.com/   ------------------------------  # Date: Fri, 01 Dec 2000 20:45:39 GMTa From: waybright@my-deja.com  Subject: Re: VMS vs. Tru64 Unix-) Message-ID: <9092lg$qr4$1@nnrp1.deja.com>S  A I should be more specific with my questions...I do appreciate the< detailed answers I've seen.l   Performance:  D Galaxy, I understand, overcomes the CPU overhead issue with VMS SMP.B The sweet spot for VMS was 12 CPU's and anything over that you hadG diminishing returns, or it did not scale linearly.  With Galaxy you can E continue to scale linearly as you add CPU's to take full advantage ofoD the GS platform.  Question:  Are there other performance benefits of? Galaxy?  I know you can can separate and tune applications morem accurately.s  E Also, does UNIX have the same CPU scaling issue that VMS had prior toV Galaxy?   . Availability - Particularly Disaster Tolerance  G VMS clustering provides the capability to have two sites separated by a,C long distance and still be part of the same cluster.  It has a wide @ variety of cluster interconnects.  If this VMS disaster tolerant clustereE is running Oracle8i, from some of the threads I've read we would need D to run Oracle Parallel server.  With Oracle Parallel Server can bothF site be PRIME, and in the event of a failure, you run a double load on5 the existing site?    I know RDB has this capability?   D Also, what are the cluster interconnects for Tru64 UNIX and what are the cluster limitations?   Benchmarks:h  E Compaq/Oracle recently announced the highest TPM-C benchmark ever foro anD Oracle DB using Tru64 UNIX.  Is there a similar benchmark for VMS or( does anyone have some performance stats?  D I read a white paper form 1998 where at the time an Alpha 8400 usingG VMS 7.1 and running Oracle 7 had the best benchmark for an Alpha serverd ever.   G Are there any whitepapers that identify Tru64 UNIX position in the UNIXt@ market?  Is Tru64 UNIX considered number one in the UNIX market?   Thanks for the help...    & Sent via Deja.com http://www.deja.com/ Before you buy.    ------------------------------   Date: 1 Dec 2000 16:47:20 -0500s9 From: Kilgallen@eisner.decus.org.nospam (Larry Kilgallen)( Subject: Re: VMS vs. Tru64 Unixd+ Message-ID: <VZ$Pjsilhx0B@eisner.decus.org>s  G In article <9092lg$qr4$1@nnrp1.deja.com>, waybright@my-deja.com writes: C > I should be more specific with my questions...I do appreciate then > detailed answers I've seen.L >  > Performance: > F > Galaxy, I understand, overcomes the CPU overhead issue with VMS SMP.D > The sweet spot for VMS was 12 CPU's and anything over that you hadI > diminishing returns, or it did not scale linearly.  With Galaxy you caniG > continue to scale linearly as you add CPU's to take full advantage of F > the GS platform.  Question:  Are there other performance benefits ofA > Galaxy?  I know you can can separate and tune applications morer
 > accurately.  > G > Also, does UNIX have the same CPU scaling issue that VMS had prior to 	 > Galaxy?n    E All SMP operating systems have such scaling issues, but the degree tomJ which they are a factor depends strongly on the nature of the application.    
 > Benchmarks:  > G > Compaq/Oracle recently announced the highest TPM-C benchmark ever fort > anF > Oracle DB using Tru64 UNIX.  Is there a similar benchmark for VMS or* > does anyone have some performance stats?    E Compaq has been unwilling to spend money for TPM-C benchmarks on VMS. C I have heard that it can cost a million dollars to assemble all thel@ hardware for such a run.  Additionally, many people scoff at theA standard benchmarks as being representative of actual application  loads.   ------------------------------  % Date: Fri, 01 Dec 2000 22:27:21 -0000-- From: wspencer@ap.nospam.org (Warren Spencer)  Subject: Re: VMS vs. Tru64 Unixp/ Message-ID: <t2g9e9572lo48e@news.supernews.com>   = waybright@my-deja.com wrote in <9092lg$qr4$1@nnrp1.deja.com>:a  
 -- snip --  H >VMS clustering provides the capability to have two sites separated by aD >long distance and still be part of the same cluster.  It has a wideA >variety of cluster interconnects.  If this VMS disaster tolerants >clusterF >is running Oracle8i, from some of the threads I've read we would needE >to run Oracle Parallel server.  With Oracle Parallel Server can both G >site be PRIME, and in the event of a failure, you run a double load on 6 >the existing site?    I know RDB has this capability?  F Yes, in Oracle Parallel server, both database instances can be active L concurrently (on different nodes in the cluster), sharing the same physical H database files.  To ensure fault tolerance, both boxes must have direct J access to the disks (don't have one box "serve" disks to the other, or it # becomes a single point of failure).0    
 -- snip --   Warren Spencer Senior Software Engineer The Associated Press  ? ** My employer does not necessarily agree with my statements **5   ------------------------------   Date: 1 Dec 2000 22:29:57 GMTc2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman) Subject: Re: VMS vs. Tru64 Unixs6 Message-ID: <9098p5$qj6$1@mailint03.im.hou.compaq.com>  G In article <9092lg$qr4$1@nnrp1.deja.com>, waybright@my-deja.com writes:iE :Galaxy, I understand, overcomes the CPU overhead issue with VMS SMP.m  >   OpenVMS Galaxy can be used for political reasons, for serverD   consolidation reasons, and to avoid bottlenecks in OpenVMS and/or B   in application code due to "non-parallelizable" code paths, codeD   paths that arise due to contention.  It also provides a high speedD   interconnect between instances -- instances can be configured intoA   an OpenVMS Cluster or can operate standalone (or combinations).O  C :The sweet spot for VMS was 12 CPU's and anything over that you hadh5 :diminishing returns, or it did not scale linearly.     E   OpenVMS SMP operates from 1 to 32 CPUs in parallel.  I've seen codefD   run with 1 CPU at most, and I've seen code running across far moreG   CPUs.  Scaling is a question of the particular job mix and the level iE   of parallelism within the job -- if you have a single-threaded job gF   (either because of code in the application or due to contention for I   an underlying mechanism), you will see no advantages from adding CPUs. -  G   See the previous discussions of Amdahl and Minsky here in comp.os.vms.D   -- this discussion was some months back -- for some discussions of   parallel processing...  F   Certain parts of I/O processing, for instance, tend to be routed to F   the primary CPU in an SMP system.  This can potentially limit systemI   I/O throughput to the performance of the primary CPU.  With Fast Path,  H   you can retarget most of the I/O processing among available secondary F   processors -- for the controllers supported with Fast Path, that is.  F :Also, does UNIX have the same CPU scaling issue that VMS had prior to :Galaxy?  D   Tru64 UNIX permits hard partitions.  OpenVMS permits hard and softD   partitions -- soft partitions (instances, in OpenVMS terminology) D   allow any secondary CPU to be shuffled over to any other instance C   without having to reboot OpenVMS.  Shuffling can occur by policy,lF   by dynamic (automatic) scheduling, or when an instance is shut down.  E :Also, what are the cluster interconnects for Tru64 UNIX and what arec :the cluster limitations?o  4   Tru64 UNIX presently supports only Memory Channel.  B :...Is there a similar benchmark for VMS or does anyone have some  :performance stats?-  &   The best benchmark is your own code.  D   If you are seriously looking at an AlphaServer GS320 series, I am H   surprised that this discussion is here in the newsgroups and not more K   directly with Compaq folks.  (Send me some information on this off-line, FH   and I can likely get somebody in more direct contact with you on this D   discussion and this comparision.)  I *think* there's a test-drive    program going...  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Fri, 01 Dec 2000 19:03:09 -0500  From: Ray <lists@aik.tec.sc.us>  Subject: Re: VMS vs. Tru64 Unixv- Message-ID: <3A283C3D.31371827@aik.tec.sc.us>    > % > VMS or Unix?  Thanks for your help!n >   K One thing to consider is that Tru64 has an 8 character username limitation.    Ray T.   ------------------------------  % Date: Fri, 01 Dec 2000 23:31:49 -0500o2 From: rdeininger@mindspring.com (Robert Deininger) Subject: Re: VMS vs. Tru64 UnixnL Message-ID: <rdeininger-0112002331500001@user-2ivebih.dialup.mindspring.com>  F In article <906ra6$1tf$1@nnrp1.deja.com>, waybright@my-deja.com wrote:  , > I'm looking for information and supporting1 > evidence for what is the preffered platform for 3 > running an Oracle 8i DB on a GS320 Alpha system -  > VMS or Tru64 Unix? > 2 > We are building a new Data base server that will0 > be the heartbeat of our business.  Needs to be/ > highly available, high performance, scalable,r2 > flexible, interoperable.  Will connect to Compaq3 > SAN.    Need to consider total cost of ownership, . > investment protection- market share, support	 > issues.e    
There is a total cost of ownership study floating around somewhere.  It compares VMS to 3 competitors' unixes  (NOT Tru64 unix) on large, mission-critical servers.  VMS came out best.  Obviously, the result will depend on what cost you assign to an hour of downtime.   I don't have a link handy, but hopefully you can find it by browsing at www.openvms.compaq.com, or maybe someone will supply a current link.   --   Robert Deininger rdeininger@mindspring.comm   ------------------------------    Date: 02 Dec 2000 12:04:13 +0800/ From: Paul Repacholi <prep@k9.prep.synonet.com>N Subject: Re: VMS vs. Tru64 Unixc0 Message-ID: <87pujbpiyq.fsf@k9.prep.synonet.com>   waybright@my-deja.com writes:>  C > I should be more specific with my questions...I do appreciate thea > detailed answers I've seen.c >  > Performance: > F > Galaxy, I understand, overcomes the CPU overhead issue with VMS SMP.D > The sweet spot for VMS was 12 CPU's and anything over that you hadI > diminishing returns, or it did not scale linearly.  With Galaxy you can3G > continue to scale linearly as you add CPU's to take full advantage of F > the GS platform.  Question:  Are there other performance benefits ofA > Galaxy?  I know you can can separate and tune applications more>
 > accurately.t  : By partitioning the total system ( almost ) you reduce the9 contention for for critical resorses. N cpus split X wayso= give N/X the contention. Well, not really, be cause the point<" of Galaxy is that SOOM are common.   > G > Also, does UNIX have the same CPU scaling issue that VMS had prior to 	 > Galaxy?/  : Worse. VMS does IO by DMA to the process. You only contend: with processes that are sharing the file. Unix does IO via> the kernel buffer cache. you are contending with ALL processes< IO ( all of it ) and memory requirements. The contention can get pretty horrid real fast.  < Also, tests I've run have shown 20:1 advantage to VMS vs DU4: in heavy paging. This was with FFTs, so don't expect it to at all repressent a DB case.  0 > Availability - Particularly Disaster Tolerance > I > VMS clustering provides the capability to have two sites separated by atE > long distance and still be part of the same cluster.  It has a widenB > variety of cluster interconnects.  If this VMS disaster tolerant	 > clustertG > is running Oracle8i, from some of the threads I've read we would need F > to run Oracle Parallel server.  With Oracle Parallel Server can bothH > site be PRIME, and in the event of a failure, you run a double load on7 > the existing site?    I know RDB has this capability?n  7 If you ARE serious about disaster tolerance, RDB is it. 4 Game set and match. Unix has no show compared to VMS	 for this.d   ~Paulr   ------------------------------  % Date: Fri, 01 Dec 2000 14:36:34 -0500-# From: Jim Agnew <agnew@hsc.vcu.edu>13 Subject: Re: Who =?iso-8859-1?Q?=B4s?= the wizard ?p* Message-ID: <3A27FDC2.89A9FDC@hsc.vcu.edu>   Steve Lionel wrote:g >    snip....   > + > The "Wizard" has multiple manifestations.o >   8 hhmm..  so, is the Wizard a hive-mind, a la Borg???  ;-)   Jima   ------------------------------  % Date: Fri, 01 Dec 2000 20:19:52 -0600e7 From: "David J. Dachtera" <djesys.nospam@earthlink.net>a3 Subject: Re: Who =?iso-8859-1?Q?=B4s?= the wizard ?n- Message-ID: <3A285C48.4FFA8214@earthlink.net>a  * fabio_compaq@ep-bc.petrobras.com.br wrote: > 1 > Who is the OpenVMS wizard ? Steve Hoffman ? :-)  > 1 > Click...  http://www.openvms.compaq.com/wizard/n   Just my opinion...  F Looks like a younger, bearded, bespectacled Jpgg (damn this keyboard!)   --   David J. Dachteras dba DJE Systemss http://www.djesys.com/  : Unofficial Affordable OpenVMS Home Page and Message Board: http://www.djesys.com/vms/soho/p  F This *IS* an OpenVMS-related newsgroup. So, a certain bias in postings is to be expected.  @ Feel free to exercise your rights of free speech and expression.  F However, attacks against individual posters, or groups of posters, are strongly discouraged.c   ------------------------------  $ Date: Fri, 1 Dec 2000 16:03:16 -0500# From: "azog" <azog@nospam.azog.org>t  Subject: Re: Who s the wizard ?& Message-ID: <3a2811f8_2@nntp2.nac.net>   I guess like the Usenet Oracle..   > >/- > > The "Wizard" has multiple manifestations.p > >  >e: > hhmm..  so, is the Wizard a hive-mind, a la Borg???  ;-) >c > Jimr   ------------------------------    Date: 02 Dec 2000 12:10:21 +0800/ From: Paul Repacholi <prep@k9.prep.synonet.com>g  Subject: Re: Who s the wizard ?0 Message-ID: <87n1efpioi.fsf@k9.prep.synonet.com>  . Steve Lionel <Steve.Lionel@compaq.com> writes:  + > The "Wizard" has multiple manifestations.o  $ NO, There is only one true wizard...  % You vanish in a puff of smoke...   ;)7   ~paull   ------------------------------  % Date: Fri, 01 Dec 2000 13:20:02 -0600i/ From: Chris Scheers <chris@applied-synergy.com> 6 Subject: Re: Why so much virtual memory for tiny jobs?O Message-ID: <7EB47F5A84E58E19.1BEB9B803396043D.0516BAD5A177BB34@lp.airnews.net>f   "Alan E. Feldman" wrote: >  > -- >  > ^^deja font fixer- >  > In articleF > <2A174DD2A44B3B2D.0E76C896AAAA890B.22DF619CA9C46C6D@lp.airnews.net>,4 >   Chris Scheers <chris@applied-synergy.com> wrote: > > "Alan E. Feldman" wrote: > > >n > > > ><J > > > > Basically, virtual pages are those pages that are reserved for theE > > > > process and might be used.  Working set pages are those pages 
 > that are! > > > > currently in real memory.s > > > >  > ; > Except for those on the free and modified lists, I think.e  F From the point of view of accounting, the free and modified lists haveE nothing to do with the process.  So they don't show up in the workings set or page file page counts.q     > > >nH > > > Or were used and might be used again! Cool. So does this mean that > A > I was referring here to "those pages that are currently in realaB > memory", not the working set pages. Just wanted to clarify that. > F > > > pages on the free list that previously belonged to a process are > alsoF > > > part of that process's virtual memory space? I guess so! (Please > > > correct me if I'm wrong.)  > >iG > > Sort of.  The term "virtual memory space" as used above isn't quitekH > > correct.  When a page is removed from the working set and put on the4 > > free list, it is no longer owned by the process. > >m > A > But aren't pages on the modified list and in the page file also D > not "owned" by the process but still in its virtual address space?H > Based on my *new* understaning, if the pages are mapped for a process,I > they are part of virtual memory (or address space) no matter where they-G > are: the working set, the free list, the modified list, the page/swapUI > files, the images, section files, what have you. Also, I thought a pages. > is "owned" only if it is in the working set.  1 Instead of "owned", substitute the word "mapped".o  A Ah, we may be getting close to the source of confusion here.  The F virtual address space is very virtual.  When you look at a page, i.e.,G read or write a memory location in the page, the page is in the workingdD set (and in physical memory).  If you are not currently looking at aC page, you don't know anything about its status.  (Sort of a twistedS3 Heisenberg Uncertainy Principle of virtual memory.)a  F Pages that are not in the working set are not mapped into the process'D virtual address space (PVAS).  The PVAS only consists of working set pages and invalid pages.  H Pages on the free or modified list, in the page file, etc., are NOT partF of the PVAS.  They may have once been part of the PVAS.  They may be aE part of the PVAS in the future, but while they are not in the working # set, they are not part of the PVAS.   D The virtual address space is really a collection of PTEs (Page TableG Entries).  When you ask if address "x" is in the virtual address space,U? you are really asking if the process has a PTE for address "x".y  F Part of the PTE indicates whether a page is valid (in the working set)B or invalid.  When a page is invalid, it is not mapped to anything.  F At attempt to access a page that doesn't have a PTE (valid or invalid) is an error (ACCVIO).h  G An attempt to access a page that has a PTE but is invalid results in anhG exception.  The exception handler examines the memory management tableslH to determine if the page can be made valid for the requested access.  IfB not, an ACCVIO error occurs.  If it can be made valid, the page isF located in memory (by reading it from its backing store (image or pageH file), locating it in the free/modified list, etc.), pointing the PTE toH the page, and making the PTE valid.  The page is then in the working set and execution can continue.i  , This process is "called faulting in a page".    H > > However, there is enough information left that if it is necessary toG > > fault the page back in, and the page is still on the free list, VMSA > canrI > > utilize the page on the free list instead of having to go back out to @ > > disk.  This is a soft fault.  A hard fault goes to the disk. > G > Hence the reason for the "secondary page cache": to try to avoid slow.6 > hard faults by replacing them with fast soft faults. >  > >[...] > E > > Keep in mind that "peak page file" and "peak virtual size" can bea > > different. > >VE > > Non-writable pages that come from images are part of the "virtuala > size"0. > > but do not take up space in the page file. > C > OK, now here you have me confused again! Check out the following:t > 
 > $ SHOW WORKa@ >   Working Set      /Limit= 895   /Quota= 1790    /Extent= 2048I >   Adjustment enabled    Authorized Quota= 1790  Authorized Extent= 2048w > $ SHOW PROC/ACCt > I > 30-NOV-2000 17:09:51.45   User: SYSTEM           Process ID:   00001A22iI >                           Node: IDS08            Process name: "_RTA1:"u >  > Accounting information:fC >  Buffered I/O count:        72  Peak working set size:        436IC >  Direct I/O count:          23  Peak virtual size:           3205 C >  Page faults:              626  Mounted volumes:                0p  >  Images activated:           5+ >  Elapsed CPU time:          0 00:00:00.18u+ >  Connect time:              0 00:00:13.45e	 > $ LO/FUy6 >   SYSTEM       logged out at 30-NOV-2000 17:10:45.63 >  >   Accounting information:uH >   Buffered I/O count:        87         Peak working set size:     436H >   Direct I/O count:          23         Peak page file size:      3205H >   Page faults:              669         Mounted volumes:             0I >   Charged CPU time:     0 00:00:00.20   Elapsed time:     0 00:01:07.63m > D > This was run on our pretty-much vanilla VMS system (v5.5-2HF). TheE > process quota WSDEFAULT is 895. The "peak working set size" is 436. A > Now, both the "peak virtual size" and "peak page file size" are0I > reported to be 3205. Now how can there be any pages in the page file ifRF > the working set size never exceeded WSDEFAULT? It was just this thatF > was confusing me in the first place! But if "peak page file size" isC > really peak virtual size, then it makes sense. For some reason, I I > previously thought that pages weren't part of virtual memory until theyaE > were at least read into physical memroy first, as I explained in myr > previous post.  4 No, virtual pages don't need to be read into memory.  G I think the quantities listed above are the number of PTEs allocated tocE the processes.  PTEs can exist for pages that were never faulted intolA memory, so the peak virtual size can be much larger than the peak  working set size.r  2 To get all the gory details, please read the I&DS.  G ----------------------------------------------------------------------- $ Chris Scheers, Applied Synergy, Inc.  C Voice: 817-237-3360            Internet: chris@applied-synergy.com n   Fax: 817-237-3074o   ------------------------------  # Date: Fri, 01 Dec 2000 22:26:47 GMT * From: Alan E. Feldman <alan48@my-deja.com>6 Subject: Re: Why so much virtual memory for tiny jobs?' Message-ID: <9098iv$q$1@nnrp1.deja.com>-  B Please do not reply to the deja address; it is broken. Post or use addresses given at the end.f  + In article <eTRipaQGcRUN@eisner.decus.org>,e    briggs@eisner.decus.org wrote:G > In article <905ov6$2ne$1@nnrp1.deja.com>, Alan E. Feldman <alan48@my-@ deja.com> writes:iF > > Or were used and might be used again! Cool. So does this mean thatD > > pages on the free list that previously belonged to a process are alsoD > > part of that process's virtual memory space? I guess so! (Please > > correct me if I'm wrong.)l >z > Well, yes and no.e > H > Virtual memory is virtual.  It isn't real.  You can't point to any oneA > real thing that say "this is part of my virtual address space".n >eG > You can say:  "Address %x7ffe0200 is within my virtual address space"s > G > This would mean that (assuming VAX hardware) that your P1LR (P1 spacesA > length register) indicated that 7ffE0200 has a page table entry  > in your page table.  >v >[...more examples omitted]U  F Okay, but getting back to the original question, What then, pray tell,C are the criteria for a page to be counted in "Virtual pages" in VMSmF statistics displays (SHOW PROCESS/ACCOUTING, SHOW PROCESS/CONTINUOUS)?G Or perhaps I should ask, in a more general form, How does VMS calculatee the number of "Virtual pages"?  F Also, do you agree that "page file size", which is used in LOGOUT/FULLF and ACC/FULL, is the same as "virtual size", which is used by the SHOWE PROC/ACC and SHOW PROC/CONT commands? My examples would lend credencef to that.   TIAf --F NOTE: If you wish to e-mail me, please do NOT use the deja address. ItE is broken. Instead, use one of the addresses below, removing the longt wrong part first. Thanks.P   Disclaimer: JMHO Alan E. Feldman  &-)+ w: afeldman@gfigroup.ButItSaidItPrinted.comc5 h: alan48@dellnet.YouCantBelieveEverythingYouRead.com     & Sent via Deja.com http://www.deja.com/ Before you buy.t   ------------------------------   End of INFO-VAX 2000.672 ************************