0 INFO-VAX	Sat, 17 Feb 2001	Volume 2001 : Issue 96      Contents:  Re: 1.2 GHz Alpha Microprocessor
 Re: Authorize  CONINTERR and friends & RE: Future outlook for OpenVMS admins?  Multitasking via telnet sessions$ Re: Multitasking via telnet sessions
 Re: newbie# Re: Now BYPASS doesn't allow DELETE  Re: OpenVMS and Supercomputing- Re: Oracle8i on OpenVMS/ES40-Platform anybody  Re: Renaming directories. How? Re: Renaming directories. How?; Re: Swapping OpenVMS & Apps from one AlphaServer to another  Re: TFT > 18.1" with 1280x1024 Re: UNIX write vs VMS ?  Re: UNIX write vs VMS ?  Re: Version numbers  Re: version numbers. RE: version numbers.? Was Renaming directories. Now: Suitable privileges for a SYSMGR C Re: Was Renaming directories. Now: Suitable privileges for a SYSMGR  which shadow patch do I need? ! Re: which shadow patch do I need?  Re: Win NT on AlphaServer 1200A  Re: www.openvms.compaq.com RE: www.openvms.compaq.com  F ----------------------------------------------------------------------    Date: 17 Feb 2001 19:16:23 +0800, From: Paul Repacholi <prep@prep.synonet.com>) Subject: Re: 1.2 GHz Alpha Microprocessor - Message-ID: <87u25tpnjs.fsf@prep.synonet.com>   5 Jeff Coffield <Jeffrey@DigitalSynergyInc.com> writes:   P > A presentaion on a new Alpha processor in a Microsoft PowerPoint presentation?  $ > Does anyone at Compaq have a clue?  F That's a rhetorical question, right? Excluding the 'usual suspects' ;)   --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.   ------------------------------  # Date: Sat, 17 Feb 2001 15:33:52 GMT $ From: Scott Vieth <svieth@wi.rr.com> Subject: Re: Authorize( Message-ID: <3A8E99EB.8090205@wi.rr.com>  , My first guess is that OPCOM ratted you out., Also, your actions were probably recorded in the system audit file.  
 -Scott :^)   William Hymen wrote:  = > Recently, I ran authorize on a test ID on a test VMS system ? > ( this system was not under security ) and reset the Password : > lifetime.  This got one of the sys admins nose all bent. > 9 > So how did he know I ran authorize on the test user ID? $ > ps- I have bypass on the test box. >  > Bill >  >    ------------------------------    Date: 17 Feb 2001 19:24:22 +0800, From: Paul Repacholi <prep@prep.synonet.com> Subject: CONINTERR and friends- Message-ID: <87pughpn6h.fsf@prep.synonet.com>   A OK, I was looking through the docs, and it seems to have vanished ? for Alpha VMS. How do I do a Connect to Interupt? Specifically, < I want to get the interupt from the *chip* console port. Not! the system console, the chip one.   A Anyone delved into this area on Alphas? Or is this stuff Vax only  nowdays?   --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.   ------------------------------  % Date: Sat, 17 Feb 2001 10:27:28 -0600 + From: "Main, Kerry" <Kerry.Main@compaq.com> / Subject: RE: Future outlook for OpenVMS admins? N Message-ID: <910612C07BCAD1119AF40000F86AF0D805284E05@kaoexc3.kao.cpqcorp.net>  I While one could split hairs and point out that the 1.5Ghz x86 is actually G slower than the 833Mhz Alpha when looking at the optimized SPECint 2000 8 result, (A=544 vs I=536) essentially, it is a dead heat.= http://ideasinternational.com/benchmark/spec/specint2000.html   L And as always, one should also consider other benchmarks and reviews as well e.g.9 <http://www4.tomshardware.com/cpu/00q4/001122/p4-04.html> 9 <http://www4.tomshardware.com/cpu/00q4/001122/p4-03.html>    Regards,  
 Kerry Main Senior Consultant  Compaq Canada Inc. Professional Services  Voice: 613-592-4660  Fax  :  819-772-7036 Email: Kerry.Main@Compaq.com     -----Original Message-----4 From: tsm@palindrome.org [mailto:tsm@palindrome.org]  Sent: February 16, 2001 11:23 AM To: Info-VAX@Mvb.Saic.Com / Subject: Re: Future outlook for OpenVMS admins?     / On 16 Feb 2001 10:30:38 +0100, Jan Vorbrueggen  8 <jan@mailhost.neuroinformatik.ruhr-uni-bochum.de> wrote:" >"rocko" <rocko1@home.com> writes: > J >> We are waiting on a couple Compaq AlphaServer ES40's to arrive.  How do they >> rate? > E >If they're the 833 MHz variety, they're the fastest processor on the  planet.   	 Not true.   I The Pentium 4 is faster than the 833 MHz EV6, at least at integer, and is 3 faster than anything but the 833 at floating point.    SpecInt2000  -----------  1.5 GHz Pentium 4 - 524  833 MHz EV6 (in ES40) - 518  1.2 GHz Athlon - 443 1 GHz Pentium III - 438  731 MHz EV6 (in ES40) - 352   
 SpecFP2000
 ---------- 833 MHz EV6 - 590  1.5 GHz Pentium 4 - 549  731 MHz EV6 - 405  1.2 GHz Athlon - 359 1 GHz Pentium III - 327    ------------------------------  # Date: Sat, 17 Feb 2001 15:46:10 GMT # From: alexlong@earthlink.net (Alex) ) Subject: Multitasking via telnet sessions 1 Message-ID: <3a8eb910.5456371@news.earthlink.net>   F I connect to a VMS server through Telnet to do various tasks, and some? of those taks take a considerable amount of time to run.  Is it B possible to multi-taks through a telnet session with just a single login?   ------------------------------    Date: 17 Feb 2001 07:58:39 -07001 From: nothome@spammers.are.scum (Malcolm Dunnett) - Subject: Re: Multitasking via telnet sessions , Message-ID: <RqKIMeH0ANEO@malvm1.mala.bc.ca>  W In article <3a8eb910.5456371@news.earthlink.net>, alexlong@earthlink.net (Alex) writes: H > I connect to a VMS server through Telnet to do various tasks, and someA > of those taks take a considerable amount of time to run.  Is it D > possible to multi-taks through a telnet session with just a single > login?  5    Sure. Depends on exactly what you're trying to do.   5    You could submit command procedures as batch jobs.   ;    You could use SPAWN/NOWAIT to run tasks in a subprocess.           ------------------------------  % Date: Sat, 17 Feb 2001 10:33:10 +0000 ) From: Christof Brass <brass@infopuls.com>  Subject: Re: newbie , Message-ID: <3A8E5366.588387C4@infopuls.com>   Andy wrote:  > G > I just started a position where i will be using open VMS extensively, J > and am wondering if this NG has any recommendations on books that covers/ > a broad range,...intro, commands, admin, etc.  > I > Any ideas appreciated,....local bookstores were not much help,..nothing 
 > in stock > 	 > thanks,  >  > andy  = While the others recommendations are all valid there could be @ something we might take into account: what is your experience so= far, what do you know? What will you do with the system? Will * you be a sysadmin, a programmer or a user?   ------------------------------  % Date: Sat, 17 Feb 2001 10:20:48 +0100   From: Paul Sture <paul@sture.ch>, Subject: Re: Now BYPASS doesn't allow DELETE+ Message-ID: <VA.000002bf.55f66499@sture.ch>   K In article <OpWi6.669$cu.2814@gazette.loc1.tandem.com>, Hoff Hoffman wrote:  > Newsgroups: comp.os.vms 4 > From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman). > Subject: Re: Now BYPASS doesn't allow DELETE% > Date: Thu, 15 Feb 2001 19:50:06 GMT  > P > In article <VA.000002a5.45845c7e@sture.ch>, Paul Sture <paul@sture.ch> writes:P > :In article <y4d7cm3h9p.fsf@mailhost.neuroinformatik.ruhr-uni-bochum.de>, Jan  > :Vorbrueggen wrote: L > :> From: Jan Vorbrueggen <jan@mailhost.neuroinformatik.ruhr-uni-bochum.de> > :> Newsgroups: comp.os.vms, > :> Subject: Re: Another missed opportunity% > :> Date: 13 Feb 2001 13:21:38 +0100  > :>  ' > :> Paul Sture <paul@sture.ch> writes:  > :>  O > :> > I once heard a tale of Pathworks files which couldn't be deleted by the  = > :> > SYTEM account - solution - reinitialize the disk. Yuk.  > :>  $ > :> I thought BYPASS bypasses that? > :>  N > :Well before I heard the tale, I thought so too. But this story came from a L > :System Manager who is one of the best I have worked with. Let's check it 	 > :out...  > L >   The system manager was wrong, or there was something else going on that K >   was mimicking a OpenVMS protection check (incorrectly).  BYPASS grants  6 >   all access to everything, even in an SEVMS system. > N Well, I did mention Pathworks:-) On further reading of the manual, I found an I explicit statement that BYPASS gives "all types of access to the object,  & regardless of its protection" (4.6.1).  P I also noted that the online version at openvms.compaq.com gave better indexing # into the manual than the SEP00 CDs.   J > :Looking at section 4.3 of the Guide to System Security, BYPASS doesn't + > :really _bypass_ access checking anymore.  > J >   The flow chart has a bug.  The READALL "no" path in HH should lead to M >   the BYPASS check, and only then to "grant" or to section II.  I've logged , >   a problem report against the flow chart. >  Thanks.  ___ 
 Paul Sture Switzerland    ------------------------------  % Date: Sat, 17 Feb 2001 10:55:17 +0000 ) From: Christof Brass <brass@infopuls.com> ' Subject: Re: OpenVMS and Supercomputing , Message-ID: <3A8E5895.F03B2291@infopuls.com>  * fabio_compaq@ep-bc.petrobras.com.br wrote: > . > Is there any supercomputer running OpenVMS ? > ? > I searched www.top500.org and I didint discover any reference 7 > about it  . . .  I dont know if  it is possible . . .  > 	 > Regards  >  > FC  ; The design, features and tools of VMS are *not* targeted to ? supercomputing where in most cases a single task for a few very = skillful users are carried out. Using VMS would be a waste of 
 resources.   ------------------------------  % Date: Sat, 17 Feb 2001 10:20:48 +0100   From: Paul Sture <paul@sture.ch>6 Subject: Re: Oracle8i on OpenVMS/ES40-Platform anybody+ Message-ID: <VA.000002be.55f66196@sture.ch>   8 In article <3A8D0C16.7EB5D8D3@home.nl>, Dirk Munk wrote:  > From: Dirk Munk <munk@home.nl> > Newsgroups: comp.os.vms 8 > Subject: Re: Oracle8i on OpenVMS/ES40-Platform anybody% > Date: Fri, 16 Feb 2001 11:16:48 GMT  >  > Hi Joseph, > J > We are in the process of converting Oracle databases form VMS 6.2/OracleH > 7.1 to VMS 7.2/Oracle 8.x. Because of an architecture change in OracleI > (this already happened in Oracle 7.3) we noticed a initial drop (!!) in E > performance of 30%. We were able to compensate this drop by several I > means. One of them was to increase the maximum size of the mailboxes to H > to the maximum 32 kB. Oracle uses a lot of buffered IO in the shape ofD > mailboxes, and increasing the size of the mailboxes reduces system > overhead.  > I Ah, bad memories of the Oracle 7.3 architecture change and an 8400 doing  G 10,000 buffered I/Os a second, ove 99.99% on mailbox reads and writes,   little real work being done.  K The DBAs sorted it out by analyzing the code. IIRC, a major recommendation  I was to switch to stored SQL procedures instead of stuff generated on the   fly.  0 Some stunning results came out of that analysis.  6 Program A. Runtime reduced from 6 hours to 20 minutes.2 Program B. Runtime reduced from 5 hours to 3 hoursE Program C. Runtime took 48 hours so we killed it. After the change 5   minutes!  ? > We also saw some poor programming techniques. Apparently some H > applications fetched their data row by row, and changing that by usingI > an array, things ran a lot quicker. (I'm not a database expert, but I'm & > sure you can figure out what I mean) > H > I don't know if the disk cabinet is already present, but we decided toH > configure our own disk cabinet. We use the 4300 shelves with 15000 rpmI > SCSI disks behind a double HSG80. These disks are so fast, that a HSG80 I > is not even capable of using their full potential. I got 35 MB/sec read I > speed on a internal shelf in the ES40, and 18 MB/sec on the HSG80 using H > a internal HSG80 performance utility. So when the HSG90 comes were are > ready :-)) > F > Because of this very high speed on the internal shelves, we will useJ > internal disks in the ES40 for scratch areas, including table spaces forG > sorting etc. Only 'real data' will be stored on the SAN. You will not G > find the 15000 rpm disks in the ES40 specs, but they are available !!  > J > I have been told by Oracle experts that Oracle advises to use big stripeF > sets (up to 256 GB) for maximum performance. I suppose this is true,G > because this way all disk in the stripe set are statistically used at I > the same IO rate. Assuming most IO is random IO, rather than sequential H > IO, this will get you maximum performance avoiding hot spots on single > disks. > H > The only save way to do this is using striped mirror sets,  = Raid 10,J > =Raid 0+1. The reads to a mirror set are distributed over all members ofF > the set, so you can again increase the IO capacity by increasing the3 > number of disks in a mirror set to more than two.  > I > And at last you can experiment with the block size as you will know. We I > haven't done any experimenting, but I tend to believe that larger block H > size will yield higher performance with today's hardware. The overheadI > of transporting more data per IO is less important than the overhead of  > more IO's. >  >  >  >  >  > joseph altea wrote:  >  > > Hi, 0 > > I am currently tasked with setting up a VLDB* > > for a compaqES40 Oracle8i. Has anybody0 > > done this type of thing out there. I am sure4 > > they have. I would like to discuss a few points. > > 9 > > In particular what kinds of strategies have you found 0 > > useful for setting up the RAID on the RA8000 > > using Oracle, etc....  > > 5 > > Any response from parties involved with Oracle on & > > OpenVMS would be most appreciated. > >  > > jaltea@dsw.net >    ___ 
 Paul Sture Switzerland    ------------------------------  % Date: Sat, 17 Feb 2001 08:30:31 +0100   From: Paul Sture <paul@sture.ch>' Subject: Re: Renaming directories. How? + Message-ID: <VA.000002b9.55916c06@sture.ch>   B In article <3A8D45BB.F8C276E0@ohio.edu>, Richard D. Piccard wrote:/ > From: "Richard D. Piccard" <piccard@ohio.edu>  > Newsgroups: comp.os.vms ) > Subject: Re: Renaming directories. How? ' > Date: Fri, 16 Feb 2001 10:22:39 -0500  > H > Glad I could help.  I think comp.os.vms is heavily populated by system > managers,     I believe you are correct there.  ? > who may well spend a large fraction of their life with BYPASS  > turned on!  M Neither necessary not safe, IMHO. Indeed I _resent_ the occasional idiot who  K sets their files to deny system access (don't they want their files backed  H up? :-) ), thereby forcing me to use BYPASS to see why a disk is full...  > > I do, but I also spend a significant part of my life helping
 > newbies. > M What has BYPASS to do with helping newbies? Use SYSPRV instead, and SET PROT  $ when you want to delete directories. >  >  > Andrew G Scott wrote:  >  > > Richard  > > C > > This was indeed the trick. I didn't look at BYPASS but used SET  > > /PROTECTION....  > > K > > All the other responses were similar in that they confirmed that RENAME L > > should do it but their default directory permissions must include DELETEL > > since no-one else mentioned that permissions would have to be reviewed.. > > 
 > > Thanks
 > > Andrew > >  > > "Richard D. Piccard" wrote:  > > K > > > RENAME is indeed the command, but it requires delete privilege on the M > > > file, and directory files are by default protected against deletion, so M > > > you have to elevate your privileges (BYPASS will certainly suffice!) or ( > > > first change the protections.  See >  > --D > ==================================================================D > Dick Piccard                           Academic Technology ManagerD > piccard@ohio.edu                                 Computer ServicesD > http://oak.cats.ohiou.edu/~piccard/                Ohio University >    ___ 
 Paul Sture Switzerland    ------------------------------  % Date: Sat, 17 Feb 2001 08:30:32 +0100   From: Paul Sture <paul@sture.ch>' Subject: Re: Renaming directories. How? + Message-ID: <VA.000002ba.55916ddd@sture.ch>   6 In article <01021612525392@lto.locktrack.com>,  wrote:" > From: l_ricker@lto.locktrack.com > Newsgroups: comp.os.vms % > Subject: Renaming directories. How?V' > Date: Fri, 16 Feb 2001 12:52:53 -07006 > 6 > > I think comp.os.vms is heavily populated by system: > > managers, who may well spend a large fraction of their > > life with BYPASS turned on!e > C > Gak!  I'd hope not, but you may be right.  Here's my own personalA > privilege discipline:  > E > a) I use a "privilege.com" command file, not SET PROCESS /PRIV=, todE >    escalate and drop specific privileges.  Benefit is that whenever F >    I "elevate" my interactive privs, the com-file sets my DCL promptD >    string to *bold* rendition, and returns it to "normal" wheneverD >    I drop them.  This way, I'm never running with BYPASS or SYSPRVE >    or whatever without an explicit visual indication of that state.eD >    Another command file, "bypass1.com", can accept one DCL commandB >    as input and execute it at temporarily elevated privilege(s). > I > b) For more extended and/or routine tasks which require elevated privs, E >    write a command file which uses F$SETPRV() to pick up *only* the-E >    higher priv's needed for a task or procedure, then reverse thesey? >    privs at com-file-exit.  The skeleton for this is roughly:D >  >    $ ! Preambles% >    $ ON CONTROL_Y THEN GOSUB Ctrl_YD >    $ ON ERROR THEN GOTO DoneB >    $ Prv = F$SETPRV("...")   !whatever elevated privs are needed > $ >    $ ... ! the task/procedure here >  >    $Done:       ! Postamble  >    $ SET NOONE= >    $ IF F$TYPE(Prv) .EQS. "STRING" THEN Prv = F$SETPRV(Prv)  >    $ EXIT  >    $ !
 >    $Ctrl_Y: $ >    $ RETURN %X2C  ! SS$_ABORT code > @ >    Note that the ON CONTROL/ERROR stmts ensure that, if I/userA >    presses <Ctrl/Y> (Interrupt), then the Postamble code always-? >    executes, thus I'm not exited still holding elevated privsMC >    without knowing it!...  YMMV, and yes, other coding techniquess9 >    and embellishments apply... this is just an outline.h > G > c) Typing the command SET PROCESS /PRIVILEGE=ALL just to "overcome" ayJ >    VMS file protection or utility's usage rule *is strongly discouraged*B >    in my shop!  So is routine login/use of the SYSTEM account... >  >    My sys-mgrs are:t > 8 >    1) trained about privs (see above, plus much more), > K >    2) equipped with individual accounts which do have SETPRV enabled, ands > K >    3) given responsibility and accountability for the use of any elevated-9 >       privs and associated tasks... consequences count.o > G >    Mostly, we get to just use common sense, but salted with the abovegL >    discipline.  We're not a "highly secure" shop, but we do take excellentK >    care of our customers'/users' data and applications.  This stuff, plusfE >    a rock-solid backup foundation and an aggressive user-management L >    discipline, remain at the core of running any successful VMS multi-user >    operation.V > K Good ideas. I really hate the tape robot utility which needs (DIAG,PHY_IO) n as it comes out of the box.   - FWIW, I've got the following in my login.com:   6 $ mypriv :== set proc/priv=('f$getjpi("","AUTHPRIV")')  M so anytime I've been running with elevated privileges I can go straight back e to my default.   ___7
 Paul Sture SwitzerlandK   ------------------------------  % Date: Sat, 17 Feb 2001 05:48:33 -0500i, From: "C Castiglia" <ccastiglia@engsint.com>D Subject: Re: Swapping OpenVMS & Apps from one AlphaServer to another+ Message-ID: <96lku3$kbb$1@bob.news.rcn.net>   L Since AlphaServer 1200's can be found in the $2,000 - $3,000 price range, itK seems to me that buying a 'standby' system with a configuration as close asaG possible to my current box will be less expensive than keeping a 4 hourc# response maintenance plan in force.a  K And, let's face it, when the stuff hits the fan and the whole company comes:6 to a standstill, even 4 hours is a long, long, time...    ? "Hoff Hoffman" <hoffman@xdelta.zko.dec.nospam> wrote in messages0 news:2Ugj6.46$fY5.387@gazette.loc1.tandem.com...A > In article <96jr2e$9gn$1@bob.news.rcn.net>, "Carmine Castiglia"-  <ccastiglia@engsint.com> writes:K > :Another question: are OpenVMS licenses tied to a specific machine or areS > :they portable?R >FG >   Check your specific licenses.  Most licenses tend to permit the usenH >   of the product on a standby system when the licensed system is down.G >   Most licenses have other requirements for more permanent transfers.p >A= > :Given two nearly identical AlphaServer 1200's, if one dieshL > :can it's hard drives be simply moved to the other machine and then booted upH > :without any significant action other than perhaps running an AUTOGEN? >a7 >   For various definitions of "nearly identical", yes.t > G >   Assuming the OpenVMS version supports the target platform, you willSH >   be able BACKUP/IMAGE and boot and use OpenVMS on it -- at worst, youG >   will have to perform a conversational bootstrap and a "USE DEFAULT"tI >   to reset the system parameters, followed by an AUTOGEN pass, followed K >   by hunting down all of the device-specific references scattered through I >   your environment (if there are configuration differences).  (Once yousH >   hunt these references a couple of times, you will hopefully start toH >   learn the value of using logical names and eschewing physical device >   names where possible.) >i, >  --------------------------- pure personal# opinion ---------------------------g1 >    Hoff (Stephen) Hoffman   OpenVMS Engineeringe hoffman#xdelta.zko.dec.com >r   ------------------------------  % Date: Sat, 17 Feb 2001 03:57:30 -0700 1 From: Vance Haemmerle <vance@toyvax.Tucson.AZ.US>v' Subject: Re: TFT > 18.1" with 1280x102451 Message-ID: <3A8DF6AA.11C8FE@toyvax.Tucson.AZ.US>l  Z > In article <3A8C65B7.D9232823@infopuls.com>, Christof Brass <brass@infopuls.com> writes:B > >I dare to ask this question because the screen will be attached > >to a VMS system.r< > >The dealer offered a Compaq 18.1" TFT display with native8 > >resolution of 1280x1024. I expect the pixels to be to? > >narrow/small to use my favourite font sizes. The dealer sayseC > >there are no bigge TFT displays available with native resolution-; > >of 1280x1024. As I have read that with TFTs it is highlyr; > >recommended to use their native resolution as they would A > >otherwise scale the picture by calculating "intermediate" dots B > >or only displaying the smaller resultion in a smaller area thanA > >the screen's size I'm seeking for a combination of screen sizes? > >and resolution which displays at the same size as my 20" CRT  > >monitor.   D   I didn't see the original post, but we use a NEC 20.1" LCD monitorC on an OpenVMS system.  It's native resolution is 1280x1024 at 60Hz.e No problems.   --B Vance Haemmerle               Internet   vance@toyvax.Tucson.AZ.USK Tucson, AZ                    Web        http://toyvax.Tucson.AZ.US/~vance/a   ------------------------------  % Date: Sat, 17 Feb 2001 03:25:45 -0800e- From: Jack Patteeuw <jjpatteeuw@peoplepc.com>S  Subject: Re: UNIX write vs VMS ?+ Message-ID: <3A8E5FB9.4AE0DB6@peoplepc.com>    Christof Brass wrote:r > A > A colleague at work is changing some C program code which callsH? > library function write() to write to an open file. He told meaA > that there are UNIX versions where this call can be interruptedl) > or fail to complete for other reasons. 0 <snip>  H He may be referring to the fact that on most Unix operating systems, the? actual write to disk is done by a separate process usually callr< "update".  It is possible if the system crashes between yourH applications write and the systems update that the data won't make it toH disk unless you are running some type of log based files system (such asG Tru64 AdvFS).  If a crash occurs on a "Unix File System" (ufs) the Unix G command fsck (file system check) is used (at boot time) to determine ifl# any repairs are required on a disk.M  
 Jack Patteeuwo   ------------------------------    Date: 17 Feb 2001 19:08:29 +0800, From: Paul Repacholi <prep@prep.synonet.com>  Subject: Re: UNIX write vs VMS ?- Message-ID: <87y9v5pnwy.fsf@prep.synonet.com>o  + Christof Brass <brass@infopuls.com> writes:    > Jan Vorbrueggen wrote:  E > > In Unix, write() is a syscall. In VMS, it is done by the C RTL, I C > > presume. In any case, VMS doesn't do such stupid things, so you F > > don't have to worry. OTOH, you can keep the loop if properly coded* > > - it will execute once, and that's it.  F > Thanks, but I will throw out the loop because on that level the codeB > can't be the same and I think it clutters the source if I have a > superflous loop.  C FOr various reasons, unix systems can 'interupt' a syscall. This ismF not an error in any normal sense of meaning, nothing is broken, exceptF unix ;), it just throws the problem back into userland and declares it 'your problem'  D VMS can do this ( or did... ) in a few odd cases. But it just pushedA back to PC, and tried again when the process was rescheduled, but' you will never see it.  C There is one case you must deal with, and that is quotas. If you doeB a write that excedes you quota, it fails with EXQUOTA. redoing the6 write, if can have the overdraft quota!, will succeed.   -- j< Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.b@                                              West Australia 6076. Raw, Cooked or Well-done, it's all half baked.   ------------------------------  % Date: Sat, 17 Feb 2001 10:20:47 +0100   From: Paul Sture <paul@sture.ch> Subject: Re: Version numbers+ Message-ID: <VA.000002bd.55f65d53@sture.ch>e  J In article <3A8D221B.55CB679B@DigitalSynergyInc.com>, Jeff Coffield wrote:5 > From: Jeff Coffield <Jeffrey@DigitalSynergyInc.com>- > Newsgroups: comp.os.vmse > Subject: Re: Version numbers' > Date: Fri, 16 Feb 2001 04:50:35 -0800j >  > JF Mezei wrote:e > N > > There was discussion about how to handle the 32k limit of version numbers. > >A8 > > My suggestion for this is to add a switch to PURGE : > >mR > > PURGE/RESET would rename files so that the lowest numbered version would be 1. > >mR > > PURGE/RESET/KEEP=0 would not delete any file, but just do the renumbering down	 > > to 0.  > >nM > > This way, when applications are started, they can use a simple command tou@ > > reset their log files to numbers that are always manageable. > >c% > > Right now, the easiest way is to:i > > 
 > > PURGE *.*3 > > RENAME *.*  ;1 > >n4 > > However you lose multiple versions of the files. >  > How about the following: >  > $       ! RENAME.COM > $       ! P1 = File name > $fL > $       HIGH_VER        = F$EXTRACT(1,5,F$PARSE(F$SEARCH(P1),,,"VERSION"))R > $       LOW_VER         = F$EXTRACT(1,5,F$PARSE(F$SEARCH(P1+";-0"),,,"VERSION")). > $       INDEX           = F$INTEGER(LOW_VER)/ > $       STOP            = F$INTEGER(HIGH_VER)i > $       COUNT           = 1d	 > $ LOOP: 6 > $       ORIG            = P1 + ";" + F$STRING(INDEX)6 > $       NEW             = P1 + ";" + F$STRING(COUNT)! > $       RENAME 'ORIG' 'NEW'/LOG % > $       INDEX           = INDEX + 1t% > $       COUNT           = COUNT + 1e+ > $       IF INDEX .LE. STOP THEN GOTO LOOP  >RO In addition to Jan's comment, on a busy system you should go back and see if a lP higher version was created in the time between you started the procedure and it 	 finished.a ___y
 Paul Sture Switzerlandt   ------------------------------  % Date: Sat, 17 Feb 2001 07:41:19 +0100r2 From: martin@radiogaga.harz.de (Martin Vorlaender) Subject: Re: version numbers.h; Message-ID: <3a8e1d0f.524144494f47414741@radiogaga.harz.de>   , Richard D. Piccard (piccard@ohio.edu) wrote:F > I suggest a new /RENUMBER option on the PURGE command as the naturalJ > enhancement.  When you are throwing away the old files is a natural time* > to decide that you want them renumbered.  D I second that. For the time being, how about the following procedure that I just wrote?   cu,n   Martin  ; $ veri = 'f$verify(f$integer("0"+f$trnlnm("reseq_verify")))- $! $! RESEQ.COM $!N $! Purge and resequence files (i.e. change their version numbers to 1,2, etc.) $! $! Parameters: $!   P1 : Filespec7 $!   P2 : maximum number of files to retain (Default 1)r $!7 $! Author: Martin Vorlaender <martin@radiogaga.harz.de>h $! $! Date: 17-Feb-2001 $! $! Disclaimer:B $!   This software is provided as-is. No guarantees are given that0 $!   it does anything useful. Use as prescribed. $.! $ on warning   then goto err_exit ! $ on error     then goto err_exit ! $ on control_y then goto err_exita $I $ Filespec  = P1 $ MaxRetain = P2 $r! $ if Filespec  .eqs. "" then exitw, $ if MaxRetain .eqs. "" then MaxRetain = "1"" $ MaxRetain = f$integer(MaxRetain) $y5 $ tmpname = "SYS$SCRATCH:RESEQ_" + f$getjpi("","pid")a $1" $ ! First purge down to n versions6 $ purge /nolog /noconfirm /keep='MaxRetain' 'Filespec' $o, $ ! Get a listing of all files to resequenceP $ directory /noheading /notrailing /columns=1 /output='tmpname'_1.TMP 'Filespec' $ C $ ! No need to go through the following if we just retain 1 versiono6 $ if MaxRetain .eq. 1 then goto skip_multiple_versions $cF $ ! Prepare dir listing for sorting: make all version numbers 5 digits, $ open /read  reseq_dirlist1 'tmpname'_1.TMP, $ open /write reseq_dirlist2 'tmpname'_2.TMP $ loop1:2 $   read /end_of_file=endloop1 reseq_dirlist1 file' $   version = f$parse(file,,,"version") J $   file = (file - version) + ";" + f$fao("!5ZL",f$integer(version - ";")). $   write /error=err_write reseq_dirlist2 file $   goto loop1 $ endloop1:. $ close reseq_dirlist2 $ close reseq_dirlist1 $h7 $ ! Resort dir listing to get ascending version numbers 3 $ lrl = f$file_attributes("''tmpname'_2.TMP","lrl")p9 $ sort /key=(character,position:1,size:'lrl',ascending) -l#     'tmpname'_2.TMP 'tmpname'_1.TMPt $d $skip_multiple_versions: $e $ ! Now do the dirty deed- $ lastfile = ""-+ $ open /read reseq_dirlist3 'tmpname'_1.TMP  $ loop2:2 $   read /end_of_file=endloop2 reseq_dirlist3 file/ $   filename = file - f$parse(file,,,"version")t $   if filename .nes. lastfile $   then $     version  = 1 $     lastfile = filenameR	 $   endifi8 $   rename /nolog /noconfirm 'file' 'filename';'version' $   version = version + 1t $   goto loop2 $ endloop2:  $ close reseq_dirlist3 $ goto cleanup $d1 $ ! Error handling (crude, but at least existant)  $ err_exit:a? $   if f$trnlnm("reseq_dirlist1","lnm$process") .nes. "" then -d         close reseq_dirlist1? $   if f$trnlnm("reseq_dirlist2","lnm$process") .nes. "" then -          close reseq_dirlist2? $   if f$trnlnm("reseq_dirlist3","lnm$process") .nes. "" then -y         close reseq_dirlist3 $- $ ! Clean up and exit 
 $ cleanup:. $   delete /nolog /noconfirm 'tmpname'_%.TMP;* $   veri = f$verify(veri)r $   exit $h $! End of RESEQ.COM    --J One OS to rule them all       | Martin Vorlaender  |  VMS & WNT programmer7 One OS to find them           | work: mv@pdv-systeme.deSJ 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, 17 Feb 2001 13:25:26 +0000 8 From: John Macallister <J.Macallister1@physics.ox.ac.uk> Subject: RE: version numbers.oL Message-ID: <35666012DF4CD411BE940090279FA240112063@ppnt41.physics.ox.ac.uk>  I > The self-re-submitting batch job log file hitting the version limit and  blowinge3 > away the job was my first exposure to this issue.   L You can avoid PURGE and RENAME problems with self-resubmitting batch jobs byL submitting the jobs from another procedure. By doing this you can manipulateL the .log files as you please. You also protect against the self-resubmittingJ batch job cycle conking out if the batch job process fails for any reason.  L The other procedure can simply be a procedure which only deals with the .logK files and then resubmits the job or you can have a single process to submit : various jobs at regular intervals or at particular times.   I I've used the single process approach myself as it's easier to coordinatemC the various regular system tasks and to balance the load over time.    John  B Name: John B. Macallister  E-mail: j.macallister1@physics.ox.ac.ukH Post: Nuclear and Astrophysics Laboratory, Keble Road, Oxford OX1 3RH,UKA Phone: +44-1865-273388 (direct)  273333 (reception)  273418 (Fax)r   ------------------------------  % Date: Sat, 17 Feb 2001 08:30:32 +0100l  From: Paul Sture <paul@sture.ch>H Subject: Was Renaming directories. Now: Suitable privileges for a SYSMGR+ Message-ID: <VA.000002bb.559170ae@sture.ch>g  F In article <3A8DE930.92527852@earthlink.net>, David J. Dachtera wrote:9 > From: "David J. Dachtera" <djesys.nospam@earthlink.net>  > Newsgroups: comp.os.vmsi) > Subject: Re: Renaming directories. How? ' > Date: Fri, 16 Feb 2001 21:00:00 -0600c >  > "Richard D. Piccard" wrote:r > > J > > Glad I could help.  I think comp.os.vms is heavily populated by systemK > > managers, who may well spend a large fraction of their life with BYPASS  > > turned on! > J > Actually, I have OPER by default and SETPRV as an authorized priv. Then,G > in my LOGIN.COM processing, I do SET PROC/PRIV=(ALL,NOBYPASS). I onlyiF > enable BYPASS to delete directories/trees or (do anything) with DFO. >lK Sorry but that's being extremely lazy, and typically leads to users asking n2 for elevated privileges which they shouldn't need.  4 I typically use the following, depending on context:  ? ALTPRI - more habit from VAX days, to control runaway processes F CMKRNL - for submit/user, run/detached, SYSGEN WRITE ACTIVE and other  useful stuff> OPER   - see operator messages; login when logins are disabled* SETPRV - get at the rest of the privileges$ SYSNAM - define system logical names, SYSPRV - only if working as a system manager2 WORLD  - monitoring and control of other processes  J GRPNAM, GRPPRV and GROUP instead of SYSNAM, SYSPRV and WORLD when working G in a group context. A pet peeve of mine is that many sites do not take e& proper advantage of the group concept.   ___u
 Paul Sture Switzerlandl   ------------------------------  % Date: Sat, 17 Feb 2001 03:29:30 -0500i- From: JF Mezei <jfmezei.spamnot@videotron.ca>.L Subject: Re: Was Renaming directories. Now: Suitable privileges for a SYSMGR, Message-ID: <3A8E365F.328AA8CE@videotron.ca>   Paul Sture wrote:s6 > I typically use the following, depending on context: > A > ALTPRI - more habit from VAX days, to control runaway processeseG > CMKRNL - for submit/user, run/detached, SYSGEN WRITE ACTIVE and other  > useful stuff@ > OPER   - see operator messages; login when logins are disabled, > SETPRV - get at the rest of the privileges& > SYSNAM - define system logical names. > SYSPRV - only if working as a system manager4 > WORLD  - monitoring and control of other processes  D In my "server" VAX, I generally manually enable SYSPRV,WORLD,SYSNAM.  S On my workstation (clustered with the first machine), I have all mighty privileges.    ------------------------------  % Date: Sat, 17 Feb 2001 08:29:38 -0500s2 From: "Richard B. Gilbert" <DRAGON@compuserve.com>& Subject: which shadow patch do I need?7 Message-ID: <200102170829_MC2-C5DC-B164@compuserve.com>e  ' Message text written by "antony wardle" * >what patch do I need to fix this problem?  4 The machines that mounted the disk is ds20e 7-2-11h1  + the machines that can't mount the disks arei   7.1 axp machines and a 6.2 vax.p   cheers   antony        JET$ mount/cluster dsa70: data_7+ %MOUNT-I-MOUNTED, DATA_7 mounted on _DSA70:-B %MOUNT-I-ISAMBR, _$1$DGA70: (METJET) is a member of the shadow set; %MOUNT-W-RMTMNTFAIL, _DSA70: failed to mount on node METMINM5 -MOUNT-F-FILESTRUCT, unsupported file structure level9; %MOUNT-W-RMTMNTFAIL, _DSA70: failed to mount on node METSATn5 -MOUNT-F-FILESTRUCT, unsupported file structure level ; %MOUNT-W-RMTMNTFAIL, _DSA70: failed to mount on node METTSTc5 -MOUNT-F-FILESTRUCT, unsupported file structure levels; %MOUNT-W-RMTMNTFAIL, _DSA70: failed to mount on node METMAXe5 -MOUNT-F-FILESTRUCT, unsupported file structure leveln; %MOUNT-W-RMTMNTFAIL, _DSA70: failed to mount on node METC01a5 -MOUNT-F-FILESTRUCT, unsupported file structure levela Jo <c  J         If, as appears likely from these messages, the disk was initializ= edJ as ODS-5, then only a VMS upgrade on the failing nodes is likely to help.=    H         VMS V7.2 made some changes to the file system; I think the majorH change was a larger BITMAP.SYS so that really big disks could have smallE cluster sizes.   If the disk in question is, say, an 18Gb disk with atJ cluster size of three, again the only solution may be to upgrade the othe= rq nodes to VMS V7.2.  F         You could, as Mark Jilson suggested, apply the appropriate VMSH Shadow patches with all their pre, co, and post requisites to all of theH systems involved.  Depending on the actual cause of the problem it mightJ help and is a good idea anyway as I think that most or all of the patches=  G involved are rated "Install_1" which could be translated as manadatory.,   ------------------------------  # Date: Sat, 17 Feb 2001 15:52:50 GMT $ From: Scott Vieth <svieth@wi.rr.com>* Subject: Re: which shadow patch do I need?& Message-ID: <3A8E9E5C.50406@wi.rr.com>   Anthony:   DO THIS:   $  maile   MAIL> send/editd To:     dsn%storage-B Subj:   pr3: problems with shadowing in mixed-architecture cluster  6 [describe your exact problem here and then hit CTRL+Z]   Get cup of coffee.   Wait for response from CSC.   H Use DSN ITS (or ftp over the Internet) to retrieve the patches mentioned in the reply from the CSC.  
 -Scott :^)   antony wardle wrote:  + > what patch do I need to fix this problem?- > 6 > The machines that mounted the disk is ds20e 7-2-11h1 > - > the machines that can't mount the disks are- > ! > 7.1 axp machines and a 6.2 vax.1 >  > cheers >  > antony >  >  > " > JET$ mount/cluster dsa70: data_7- > %MOUNT-I-MOUNTED, DATA_7 mounted on _DSA70: D > %MOUNT-I-ISAMBR, _$1$DGA70: (METJET) is a member of the shadow set= > %MOUNT-W-RMTMNTFAIL, _DSA70: failed to mount on node METMIN 7 > -MOUNT-F-FILESTRUCT, unsupported file structure level = > %MOUNT-W-RMTMNTFAIL, _DSA70: failed to mount on node METSAT17 > -MOUNT-F-FILESTRUCT, unsupported file structure level = > %MOUNT-W-RMTMNTFAIL, _DSA70: failed to mount on node METTSTt7 > -MOUNT-F-FILESTRUCT, unsupported file structure level = > %MOUNT-W-RMTMNTFAIL, _DSA70: failed to mount on node METMAX 7 > -MOUNT-F-FILESTRUCT, unsupported file structure levelt= > %MOUNT-W-RMTMNTFAIL, _DSA70: failed to mount on node METC01e7 > -MOUNT-F-FILESTRUCT, unsupported file structure leveln > Js >  >    ------------------------------  % Date: Sat, 17 Feb 2001 10:53:27 +0000i) From: Christof Brass <brass@infopuls.com>d( Subject: Re: Win NT on AlphaServer 1200A, Message-ID: <3A8E5827.FA9478AE@infopuls.com>   Carmine Castiglia wrote: > M > I am considering setting up an AlphaServer 1200A as a Win NT server on a 30xL > user LAN.  Any comments on the latest version of Win NT or Win 2000 ServerE > which would be supported would be appreciated, as well as any other: > concerns/comments.  @ Don't use Micro$oft products as file servers. Obviously from you@ post I don't know what you're planing to use this server for but; file serving seems a reasonable cause. BTW this topic isn'tB  related to *this* NG in any way.   ------------------------------  % Date: Sat, 17 Feb 2001 08:30:33 +0100a  From: Paul Sture <paul@sture.ch># Subject: Re: www.openvms.compaq.com + Message-ID: <VA.000002bc.5591723e@sture.ch>l  < In article <3A8C2EB6.6C07A4E3@gmx.ch>, Didier Morandi wrote:. > From: Didier Morandi <Didier.Morandi@gmx.ch> > Newsgroups: comp.os.vmst% > Subject: Re: www.openvms.compaq.comb' > Date: Thu, 15 Feb 2001 20:32:06 +0100t >  > Anecdote:l > I > Back in the early-80s, when DASSAULT Aircrafts in Paris decided to move H > from a good ol' PDP 11-44 RSX11M Basic+2 to a VAX 750 VMS 2.5 (the oneH > with which you were able to delete non-empty directories :-), I copiedJ > form the PDP a _huge_ Basic program which took something like 45 minutesI > to be compiled/TKBzed, then I decided to compile/link it on the VAX. ItPI > was 11:55 am. I said: "Pals, let's start that compile/link during lunch8H > time and we will see later". I hit return then stood up from my place.I > Suddenly I got back my prompt. It was _finished_, and no errors (thanks<E > to the RSX1 AME folks by the way...) So my pals said. "ok, you stay, > here, we go for lunch".M > , I trust you joined them for lunch anyway :-)   Early 80's memories of Paris:d    Picture a seething manager type.   "Where the *%&# have you been?"i "Lunch"y "Oh, sorry. Which restaurant?"  I The next 15-30 minutes then spent discussing the restaurant, what was on  4 the menu, which wine we had and who chose it, etc...   Fond memories... ___e
 Paul Sture Switzerland    ------------------------------  % Date: Sat, 17 Feb 2001 08:25:50 -0500n From: stan@stanq.com# Subject: RE: www.openvms.compaq.comx. Message-ID: <3A8E358E.26022.513DD71@localhost>  - On 16 Feb 2001, at 21:18, Hoff Hoffman wrote:M= >   For best flavor, please swap the underscore for a hyphen:h > : >     http://www.openvms.compaq.com/misc/about-server.html  B Hmmm.  That page has "Powered by Purveyor" and "Java by OpenCube" @ logos, but no "Powered by OpenVMS" logo.  Maybe Compaq needs to " develop one that we can all use...     --Stan  
 ----------G Stanley F. Quayle, P.E.   N8SQ   +1 614-868-1363   Fax: +1 614 868-1671 1 8572 North Spring Ct. NW, Pickerington, OH  43147 = Preferred address:  stan@stanq.com       http://www.stanq.com    ------------------------------   End of INFO-VAX 2001.096 ************************