1 INFO-VAX	Thu, 06 Nov 2003	Volume 2003 : Issue 615       Contents: Re: (Open)SSL for VMS  Re: Announcing the FutureVAX0 Re: concerned and confused about MOUNT/NOREBUILD. Configuration Question TCP/IP Services for VMS. Re: Daylight to Standard Time Change via Batch. Re: Daylight to Standard Time Change via Batch. Re: Daylight to Standard Time Change via Batch. Re: Daylight to Standard Time Change via Batch. Re: Daylight to Standard Time Change via Batch. Re: Daylight to Standard Time Change via Batch. Re: Daylight to Standard Time Change via Batch. Re: Daylight to Standard Time Change via Batch: Re: Does anyone shut down for system disk backup any more?: Re: Does anyone shut down for system disk backup any more?: Re: Does anyone shut down for system disk backup any more?: Re: Does anyone shut down for system disk backup any more?: Re: Does anyone shut down for system disk backup any more?: Re: Does anyone shut down for system disk backup any more?: Re: Does anyone shut down for system disk backup any more?: Re: Does anyone shut down for system disk backup any more?: Re: Does anyone shut down for system disk backup any more?: Re: Does anyone shut down for system disk backup any more?: Re: Does anyone shut down for system disk backup any more? Re: Erp running on Vms ?# Expected transfer speeds using DSSI  Re: Fortran calling sequence Re: Full Duplex NIC cards  Re: Full Duplex NIC cards  Re: Help W/NAS proposal... Re: Help W/NAS proposal...A Re: In CSWS (Apache) CGI, what's the IP address of the client (br . Looking for a disk dump over TCP/IP connection! Re: newbie problem installing VMS ! Re: newbie problem installing VMS * Opinions on Nemonix Ultra SCSI XMI adapter0 Re: Piping X-Windows traffic over ssh on OpenVMS$ Questions on VERIFY at process start Re: SAN Switch Issue RE: SAN Switch Issue Re: SAN Switch Issue- Re: To foreign or verb ? That is the question 9 VAX/VMS to Itanium Migration News, November issue is here * Re: VMS in pole position in an HP WEB page$ Re: VMS not mentioned by HP....again Re: VUP? Re: VUP? Re: VUP? Re: VUP? Re: VUP? [OT] - VAX9000 (was: Re: VUP?)" RE: [OT] - VAX9000 (was: Re: VUP?)? [semi-OT] SCSI jumpers (was: Re: newbie problem installing VMS) C Re: [semi-OT] SCSI jumpers (was: Re: newbie problem installing VMS)   F ----------------------------------------------------------------------  % Date: Wed, 05 Nov 2003 21:33:28 +0100  From: Dirk Munk <munk@home.nl> Subject: Re: (Open)SSL for VMS2 Message-ID: <bobn0m$b6e$1@news2.tilbu1.nb.home.nl>   Jakob Erber wrote: > Hello, > L > as far as I know, there is a official SSH package now from HP for OpenVMS.N > Does this mean, there is also an official SSL module? Or is it still OpenSSLI > to be used? If the latter is true, where can I find the newest version?  > / > sorry, if this question has been asked before  >  > best regards >  > Jakob  >  > Q The OpenSSL package from HP will be used with TCPIP V5.4, if you want to use pop  Q mail with SSL etc. (startup SSL first, and TCPIP later). I suppose this means it  H will be a more integral part of VMS instead of 'just a nice goodie' :-).   ------------------------------  % Date: Wed, 05 Nov 2003 23:36:58 +0000 ) From: Antonio Carlini <arcarlini@iee.org> % Subject: Re: Announcing the FutureVAX > Message-ID: <tQfqb.3583$382.824@newsfep4-winn.server.ntli.net>   Fabio Cardoso wrote:  A > This is a nice idea but I dont understand why HP/Compaq through @ > the Services Group didnt developed a project like this before.9 > They could seel Proliant DL-360, Windows XP Server and  . > Charon-VAX to substitute the obsolete VAXes.  ; Isn't Charon-VAX developed by a bunch of people who used to 9 work for DEC? There are plenty of new companies out there : created by the fallout from the Compaq takeover of DEC and by the HP takeover of Compaq.   > So HP/Compaq didn't want to develop this market for themselves4 and were happy for someone else to go off and do it.  - At least that's what it looks like from here.   = You'll note that the licencing appears to require HP hardware 9 (or maybe you just need that for support ... but not many 2 enterprise-level customers will risk doing without	 support!)    Antonio    --     --   --------------- - Antonio Carlini             arcarlini@iee.org    ------------------------------  % Date: Thu, 06 Nov 2003 02:43:58 +0800 , From: Paul Repacholi <prep@prep.synonet.com>9 Subject: Re: concerned and confused about MOUNT/NOREBUILD - Message-ID: <87isly1wr5.fsf@prep.synonet.com>   < "David McKenzie" <david.mckenzie@paradigm-shift.biz> writes:  B > Rebuild is not a big deal. The operating system allocates blocksE > from the free list to itself so it can avoid writing the bit map at F > write time. All you lose by not rebuilding is space, and givn modern > disks, not much of that.   Rephrasing:   < Each node builds a private list of free blocks so they don'tB constantly hammer on the bitmap file. Blocks that are in the nodesD freelist are marked as in use in the bit map. These blocks come fromF delete/truncates on the node, with the freed blocks being added to the@ free block list if it is smaller than the target size. This also+ reduces the bitmap accesses on deletes btw.   ; > However at some stage /rebuild has to take out the volume D > serialization lock so at that point all other io stops to the disk  A IO can continue, but allocation/dealocations are stalled till the  allocation lock is released.  9 > Rebuild some time after reboot when the system is quiet   @ Note that quiet means IO and allocation/dealocs, rather than CPU
 or memory.   --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076* comp.os.vms,- The Older, Grumpier Slashdot. Raw, Cooked or Well-done, it's all half baked.F EPIC, The Architecture of the future, always has been, always will be.   ------------------------------   Date: 5 Nov 2003 22:45:17 -0800 + From: tdreher@economweb.com (Thomas Dreher) 7 Subject: Configuration Question TCP/IP Services for VMS = Message-ID: <fa2e75be.0311052245.39d7cd76@posting.google.com>   C Hi, I would like to add an AlphaStation 200 named "speedy" running  D OpenVMS 7.2-1 and TCP/IP Services for VMS V5.0A to my local network.C The local network consists of 5 PCs, all connected through a hub to A an ADSL Modem/Router. The Alpha Box is also connected to the hub.   ? The PCs on the network getting their IP Addresses from the ADSL 
 Modem/Routers 
 DHCP Service.   
 Questions:  A a) can I configure the TCP/IP Services for VMS such, that it will 	 obtain an C    IP Address from the ADSL Modem/Routers DHCP Service, and if yes,  how?  > b) I would be happy to set a static IP Address for the Alpha, F    I just need to be able to have a RPC connection between the PCs and the 
    Alpha.   D What I did so far is adding the Alpha's name and the Routers Name in the host database.1 like:  TCPIP>set host speedy /address=192.168.0.6  /network_submask=255.255.255.04        TCPIP>set host billion /address=192.168.1.254  > When I try to Ping to the routers address 192.168.1.254 is not0 successful and responds with "no route to host".  A Can somebody help me and tell me, what else has to be configured.  -Routing? Name Server, etc?   " I would be grateful for any hints.   thansk and regards   Thomas   ------------------------------  # Date: Wed, 05 Nov 2003 19:31:53 GMT 3 From: hammond@not@peek.ssr.hp.com (Charlie Hammond) 7 Subject: Re: Daylight to Standard Time Change via Batch 2 Message-ID: <Jecqb.8489$LQ5.2826@news.cpqcorp.net>   Sombody wrote:  B >>SYS$EXAMPLES:DAYLIGHT_SAVINGS.COM invokes UTC$CONFIGURE_TDF.COM. > L >Only on OpenVMS VAX, and on OpenVMS Alpha < V7.3. On OpenVMS Alpha V7.3 andA >higher, DAYLIGHT_SAVINGS.COM invokes/submits UTC$TIME_SETUP.COM.   I On OpenVMS Alpha V7.3 and higher, and on all OpenVMS I64 systems you will G find that UTC$CONFIGURE_TDF.COM and UTC$TIMEZONE_SETUP.COM exit with a  
 warning that:   C       This procedure is obsolete and does not produce fully correct @       results in all cases.  To setup all time zone information,+       please execute the following command:   +           $ @SYS$MANAGER:UTC$TIME_SETUP.COM   I UTC$CONFIGURE_TDF.COM and UTC$TIMEZONE_SETUP.COM are not being maintained   and they will not be maintained.  E The original code is still there after the $EXIT for those that care, E but I WOULD NOT suggest using it.  The comments in the new version of D UTC$TIME_SETUP.COM explain how to perform functions similar to thoseI previously performed by UTC$CONFIGURE_TDF.COM and UTC$TIMEZONE_SETUP.COM.    --J       Charlie Hammond -- Hewlett-Packard Company -- Ft Lauderdale  FL  USAF           (hammond@not@peek.ssr.hp.com -- remove "@not" when replying)J       All opinions expressed are my own and not necessarily my employer's.   ------------------------------  # Date: Wed, 05 Nov 2003 20:20:17 GMT 6 From: peter@langstoeger.at (Peter 'EPLAN' LANGSTOEGER)7 Subject: Re: Daylight to Standard Time Change via Batch 5 Message-ID: <5Ycqb.202798$vL1.2411588@news.chello.at>   \ In article <03102722323988@dscis6-0.dalsemi.com>, brandon@dalsemi.com (John Brandon) writes: >Carl Karcher wrote E >> Consider setting sysgen parameter AUTO_DLIGHT_SAV=1 on your 7.3 or C >> later systems (must be the same on all members of a cluster). It E >> does work as advertised and you can forget about all your previous  >> procedures. > # >Well I wish it was that easy Carl!  > I >In our application environment can not tolerate a snap backwards in time O >service. It has the potential to cause dupliate records in our RMS ISAM files.   F Then why not using the UTC (with the local time and the system logical; SYS$TIMEZONE_DIFFERENTIAL it should be easily computable) ?   O >Too bad there is not an AUTO_DLIGHT_SAV=2 where this option skews the clock by F >the combination of AUTO_DLIGHT_TIMEADJUST and AUTO_DLIGHT_TICKLENGTH.   Submit an SPR. Maybe for V9    --   Peter "EPLAN" LANGSTOEGER % Network and OpenVMS system specialist  E-mail  peter@langstoeger.atF A-1030 VIENNA  AUSTRIA              I'm not a pessimist, I'm a realist   ------------------------------   Date: 5 Nov 2003 15:14:18 -0600 4 From: kuhrt@nospammy.encompasserve.org (Marty Kuhrt)7 Subject: Re: Daylight to Standard Time Change via Batch 3 Message-ID: <pjOefZC88MQm@eisner.encompasserve.org>   h In article <y67qb.8440$Tf5.7745@news.cpqcorp.net>, hammond@not@peek.ssr.hp.com (Charlie Hammond) writes:N > Although this suggestion does not address the issue of existing applciationsN > which may not be practical to change, the issue of transaction serializationG > might be better addressed by using GMT or UCT rather than local time. ? > This avoids the daylight/standard time change issue entierly.   @ I tried to use the gmtime calls in C and they relied on the UTC A settings, especially SYS$TIMEZONE_DIFFERENTIAL.  Since it is this A logical that I had problems with in the first place, so using GMT 7 was no gain.  At least that is the way it seemed to me.    ------------------------------  # Date: Wed, 05 Nov 2003 20:58:51 GMT # From: hoff@hp.nospam (Hoff Hoffman) 7 Subject: Re: Daylight to Standard Time Change via Batch 2 Message-ID: <fwdqb.8507$b%5.7538@news.cpqcorp.net>  n In article <5Ycqb.202798$vL1.2411588@news.chello.at>, peter@langstoeger.at (Peter 'EPLAN' LANGSTOEGER) writes:] :In article <03102722323988@dscis6-0.dalsemi.com>, brandon@dalsemi.com (John Brandon) writes: J :>In our application environment can not tolerate a snap backwards in timeP :>service. It has the potential to cause dupliate records in our RMS ISAM files. : G :Then why not using the UTC (with the local time and the system logical < :SYS$TIMEZONE_DIFFERENTIAL it should be easily computable) ? : P :>Too bad there is not an AUTO_DLIGHT_SAV=2 where this option skews the clock byG :>the combination of AUTO_DLIGHT_TIMEADJUST and AUTO_DLIGHT_TICKLENGTH.  :  :Submit an SPR. Maybe for V9    J   This request comes up occasionally -- drifting the clock does look nice,J   but it inot what I would consider standard and expected behaviour of theI   timekeeping around the daylight savings time switch-over.  Accepted and K   expected practice is the existing one-hour change -- anything approaching J   the drift would have to be site-specific.  (A one-hour delta for the DSTH   drift is also a substantial drift, too.  NTP doesn't like to see delta%   values this large.  But I digress.)   F   UTC would be the obvious choice, either directly through applicationD   modifications or by setting the system time to UTC (or to standardG   time) and leaving it there.  Alternatively or additionally, you could E   code the application to use the DST-change notification events that "   are available in V7.3 and later.  F   Were it to be designed anew, there would almost certainly be changesD   made within OpenVMS timekeeping, and the core hardware clock wouldE   probably be set to UTC.  As it stands, changes within this code are C   not trivial, given compatibility expectations -- what needs to be G   done is arguably simple, but the implementation and the relationships    get interesting.  F   Extending this, another proposal that was in common circulation backE   around Y2K involved the process-local time, and that builds on what @   was discussed here, adding yet whole new layers of "interest".  E   (This timekeeping stuff is deceptively difficult when compatibility    is an expectation.)     N  ---------------------------- #include <rtfaq.h> -----------------------------K     For additional, please see the OpenVMS FAQ -- www.hp.com/go/openvms/faq N  --------------------------- pure personal opinion ---------------------------E         Hoff (Stephen) Hoffman   OpenVMS Engineering   hoff[at]hp.com    ------------------------------  % Date: Wed, 05 Nov 2003 16:41:08 -0500 * From: JF Mezei <jfmezei.spamnot@istop.com>7 Subject: Re: Daylight to Standard Time Change via Batch ) Message-ID: <3FA96E6E.AFAC89F9@istop.com>    Hoff Hoffman wrote: G >   (This timekeeping stuff is deceptively difficult when compatibility  >   is an expectation.)   J In terms of applications, how many sources of time are there ? Do all timeB services end up calling $GETTIM to get current time ? Or do enoughK applications cheat and go fetch some value in a specific memory location to  get current time ?   ------------------------------  % Date: Wed, 05 Nov 2003 19:50:26 -0600 1 From: "David J. Dachtera" <djesys.nospam@fsi.net> 7 Subject: Re: Daylight to Standard Time Change via Batch ' Message-ID: <3FA9A8E2.A3E9CE0E@fsi.net>    Charlie Hammond wrote: > N > Although this suggestion does not address the issue of existing applciationsN > which may not be practical to change, the issue of transaction serializationG > might be better addressed by using GMT or UCT rather than local time. ? > This avoids the daylight/standard time change issue entierly.  > F > Even better is to use an application controlled transaction counter,E > which avoids porblems from time changes for any reason.  (Digital's B > DBMS and RDB datasbase management systems used such a counter toF > serialize journal records used for recovery and rollback.  I presume1 > the ORACLE version of these producst still do.)   H The DBMS is not necessarily the issue. The issue in my case is date/timeG stamps for medical records. It'd look kind of bizarre to have a patient H with a gallbladder removed even before being taken into surgery, or have- a baby born before mom went into labor, or...   H There are more practical examples relating to questions of liability, of course, but...   --   David J. Dachtera  dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------  % Date: Wed, 05 Nov 2003 19:54:15 -0600 1 From: "David J. Dachtera" <djesys.nospam@fsi.net> 7 Subject: Re: Daylight to Standard Time Change via Batch ' Message-ID: <3FA9A9C7.ECC92E34@fsi.net>     Peter 'EPLAN' LANGSTOEGER wrote: > ^ > In article <03102722323988@dscis6-0.dalsemi.com>, brandon@dalsemi.com (John Brandon) writes: > >Carl Karcher wrote G > >> Consider setting sysgen parameter AUTO_DLIGHT_SAV=1 on your 7.3 or E > >> later systems (must be the same on all members of a cluster). It G > >> does work as advertised and you can forget about all your previous  > >> procedures. > > % > >Well I wish it was that easy Carl!  > > K > >In our application environment can not tolerate a snap backwards in time Q > >service. It has the potential to cause dupliate records in our RMS ISAM files.  > H > Then why not using the UTC (with the local time and the system logical= > SYS$TIMEZONE_DIFFERENTIAL it should be easily computable) ?   C If you can sell the hospital on the idea of having the ENTIRE staff H (doctors, nurses, etc.) trained to deal with with UTC in 24 hour format,
 go for it!   --   David J. Dachtera  dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------  % Date: Wed, 05 Nov 2003 22:06:07 -0500 * From: JF Mezei <jfmezei.spamnot@istop.com>7 Subject: Re: Daylight to Standard Time Change via Batch ) Message-ID: <3FA9BA97.5C27E1EB@istop.com>    "David J. Dachtera" wrote:J > The DBMS is not necessarily the issue. The issue in my case is date/timeI > stamps for medical records. It'd look kind of bizarre to have a patient J > with a gallbladder removed even before being taken into surgery, or have/ > a baby born before mom went into labor, or...   G If you watch enough Star Trek, these things become quite commonplace...   H Consider airline schedules. They really need to handle time differences.I Consider a flight between north america and australia where australia and K n-america not only have different dates but also opposite time changes (one L moves ahead while the other moves back) and don't even change their times at the same time.  H Having one fixed time helps a lot. And that is GMT. (England is at GMT+1 during summer BTW).   G Having said this, having the time stamp in GMT does present interesting N problem. When you move the time back one hour you not only need to ensure thatM the logical which defines this node's GMT offset is changed at the same time, I but also that any application that has translated that logical is told to 7 retranslate the logical because the offset had changed.     M It might be simpler to have the clock set to GMT (never needing time changes) G and then have applications add the GMT offset when displaying the time, # according to the local node's time.    ------------------------------  % Date: Wed, 05 Nov 2003 13:11:29 -0600 ( From: brandon@dalsemi.com (John Brandon)C Subject: Re: Does anyone shut down for system disk backup any more? 1 Message-ID: <03110513112975@dscis6-0.dalsemi.com>   M > >> I can think of 2 VMS-supplied  .EXE files that ARE modified, off the top  > >> of my head. > > M > > Which ones?  I was not aware of this.  Why?  That seems a bit nuts to me.  > > I >    DCLTABLES.EXE gets modified by many software installations to add or  >    update a command. > K >    SYS.EXE gets modified every time you do a "set time=xxx" command, such E >    as during a normal system shutdown.  That's where time time cell I >    exists that gets checked against the battery powered hardware clock  K >    during boot to determine whether VMS should use the hardware clock or   >    prompt for the time.  > K >    Some layered products maintain data in global sections backed by .EXE   >    files.  > @ >    Howver, generally neither of these should be a problem for  >    /ignore=backup.  N The DCLTABLES.EXE would typically not be a problem then - unless of course youH would be performing an application install during your /IGNORE=INTERLOCK backup.   L The SYS.EXE - that makes sense, however I am not in the habit of using the  L "SET TIME=xxx " during my backups either.  I assume that NTP does not affectL this since it is dwiddles the EXE$GL_TIMEADJUST & EXE$GL_TICKLENGTH internal
 variables.  3 Global sections - agreed.  No problem there either.    Thanks!        J*o*h*n B*r*a*n*d*o*n  VMS Systems Administrator * firstname.lastname.spam.me.not@dalsemi.com   ------------------------------  % Date: Thu, 06 Nov 2003 02:06:12 +0800 , From: Paul Repacholi <prep@prep.synonet.com>C Subject: Re: Does anyone shut down for system disk backup any more? - Message-ID: <877k2e3d2j.fsf@prep.synonet.com>   7 rdeininger@mindspringdot.com (Robert Deininger) writes:   E > I can think of 2 VMS-supplied .EXE files that ARE modified, off the  > top of my head.    What is the other one?     --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076* comp.os.vms,- The Older, Grumpier Slashdot. Raw, Cooked or Well-done, it's all half baked.F EPIC, The Architecture of the future, always has been, always will be.   ------------------------------  % Date: Thu, 06 Nov 2003 02:15:48 +0800 , From: Paul Repacholi <prep@prep.synonet.com>C Subject: Re: Does anyone shut down for system disk backup any more? - Message-ID: <87y8uu1y23.fsf@prep.synonet.com>   K Not from John, but the line eater seems to have snacked on the parent post:   A >> An interesting experiment would be to find an old disk with an D >> easily accessible "write protect" button on the outside.  Boot upB >> VMS from such a disk, get your applications going (yes, even ifD >> they are on a different disk), and then write protect your systemA >> disk.  I wonder how many components would start failing and/or  >> complaining?   B Or consider how you now install VMS. Does it have a read-only lookA about it? Now compare the parameters to a normal running system's  ones.    --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076* comp.os.vms,- The Older, Grumpier Slashdot. Raw, Cooked or Well-done, it's all half baked.F EPIC, The Architecture of the future, always has been, always will be.   ------------------------------  % Date: Thu, 06 Nov 2003 02:11:01 +0800 , From: Paul Repacholi <prep@prep.synonet.com>C Subject: Re: Does anyone shut down for system disk backup any more? - Message-ID: <873cd23cui.fsf@prep.synonet.com>   , JF Mezei <jfmezei.spamnot@istop.com> writes:   > Robert Deininger wrote:   E >> Are there additional files that don't show up in SHOW DEVICE/FILE, ! >> but which are actually "open"?   < > If that is the case, it would be a serious flaw in VMS andC > SHOW/DEV/"FILE. One must howevbere be careful about sporadic file D > access (open file, read record, close file) which would not likely5 > show up at the time SHOW DEV/FILES would be issued.   A It has been the case for ages, if not forever with VMS. These are E files that are opened by the primative file system to boot the system ! before the XQP is up and running.    --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076* comp.os.vms,- The Older, Grumpier Slashdot. Raw, Cooked or Well-done, it's all half baked.F EPIC, The Architecture of the future, always has been, always will be.   ------------------------------  % Date: Wed, 05 Nov 2003 13:18:45 -0700 + From: "Barry Treahy, Jr." <Treahy@MMaz.com> C Subject: Re: Does anyone shut down for system disk backup any more? ' Message-ID: <3FA95B25.7050707@MMaz.com>    Bob Koehler wrote:  ] >In article <03110508155559@dscis6-0.dalsemi.com>, brandon@dalsemi.com (John Brandon) writes:  >    > K >>>I can think of 2 VMS-supplied  .EXE files that ARE modified, off the top  >>>of my head.	 >>>        >>> K >>Which ones?  I was not aware of this.  Why?  That seems a bit nuts to me.  >> >>     >> > H >   DCLTABLES.EXE gets modified by many software installations to add or >   update a command.  >    > D Yes, but then you must reinstall the new image, otherwise the older G image will still be the active CLI when processes are created.  If you  G do not, then the new table will not have any problems being backed up,   because it will not be open.    J >   SYS.EXE gets modified every time you do a "set time=xxx" command, suchD >   as during a normal system shutdown.  That's where time time cellH >   exists that gets checked against the battery powered hardware clock J >   during boot to determine whether VMS should use the hardware clock or  >   prompt for the time. >  >    > F Nope, at least not on 7.2.  I just made a copy of the image, issued a E set time altering the system time, and then made another copy of the   image and itis identical.   J >   Some layered products maintain data in global sections backed by .EXE 
 >   files. >    > I That is never meant to be static, when the global sections are unmapped,   the data is gone!   ? >   Howver, generally neither of these should be a problem for   >   /ignore=backup.  >  >  >    >  Barry    --    > Barry Treahy, Jr                       E-mail: Treahy@MMaz.com> Midwest Microwave                          Phone: 480/314-1320> Vice President & CIO                         FAX: 480/661-7028                            ------------------------------  % Date: Wed, 05 Nov 2003 15:31:50 -0500 * From: JF Mezei <jfmezei.spamnot@istop.com>C Subject: Re: Does anyone shut down for system disk backup any more? ) Message-ID: <3FA95E35.5222187D@istop.com>   C > >> An interesting experiment would be to find an old disk with an F > >> easily accessible "write protect" button on the outside.  Boot upD > >> VMS from such a disk, get your applications going (yes, even ifF > >> they are on a different disk), and then write protect your system
 > >> disk.    M The audit server would get mighty mad at not being able to write its logs and M probably freeze the system. OPCOM woudl also get mighty mad with inability to N write to operator.log, and the queue manager would also complain bitterly. You  woudln't have a happy family :-)   ------------------------------  % Date: Wed, 05 Nov 2003 15:56:39 -0600 ( From: brandon@dalsemi.com (John Brandon)C Subject: Re: Does anyone shut down for system disk backup any more? 1 Message-ID: <03110515563953@dscis6-0.dalsemi.com>   E > > >> An interesting experiment would be to find an old disk with an H > > >> easily accessible "write protect" button on the outside.  Boot upF > > >> VMS from such a disk, get your applications going (yes, even ifH > > >> they are on a different disk), and then write protect your system > > >> disk. > O > The audit server would get mighty mad at not being able to write its logs and O > probably freeze the system. OPCOM woudl also get mighty mad with inability to P > write to operator.log, and the queue manager would also complain bitterly. You" > woudln't have a happy family :-)  4 All of which these can be moved off the system disk.  4 Please refer to the following WIS (DSNlink) article:= TITLE: [OpenVMS] Methods Used To Recover Space On System Disk   L This also caught my eye - however have not had time to review - looks like a
 good read:  F TITLE: VMS V5.n: How Can A Layered Product Be Removed From The System?         J*o*h*n B*r*a*n*d*o*n  VMS Systems Administrator * firstname.lastname.spam.me.not@dalsemi.com   ------------------------------   Date: 5 Nov 2003 19:57 CST. From: carl@nospam.gerg.tamu.edu (Carl Perkins)C Subject: Re: Does anyone shut down for system disk backup any more? 3 Message-ID: <5NOV200319571074@nospam.gerg.tamu.edu>   , brandon@dalsemi.com (John Brandon) writes...M }> As most others have mentioned, standalone is the safest, but generally not ' }> an available option on most systems.  } 	 }I agree!   ; I'm not sure about "most" - certainly "many", maybe "most".   M }> The one thing that might be worth your while is doing a separate backup of N }> the queue database to a safe spot prior to your on-line image backup of theN }> disk.  We use a routine that does a CONVERT/SHARE of the files to a supportN }> directory on another disk, from which we could restore if it were necessaryA }> (I think this came from an old DSNlink article circa VAX v5.5)  } P }I assume that you maintain your queue database on the system disk, why not moveO }it to a non-system disk using the QMAN$MASTER logical to define that location?  } M }> That having been said, I have restored a few systems from on-line backups, M }> and only needed to re-create the queue files once.  Usually just needed to   }> re-populate the batch entries } 7 }That would eliminate the re-create as mentioned above.  } N }My policy - if I can define a logical and subsequently move a file off of the }system disk - I am so there!  }  }  }J*o*h*n B*r*a*n*d*o*n }VMS Systems Administrator+ }firstname.lastname.spam.me.not@dalsemi.com   B I do not beleive that this solves the problem, it just moves it to
 another disk.   E Instead of the file maybe being bad on the the system disk backup, it B is maybe bad on that other disk's backup. I do not see how this isC significantly better. The only way this can help is if that disk is D backed up a lot more often than the systme disk, giving you a chanceD to pull a good copy off of an earlier backup - but that is certainlyG not usually going to be better than restoring it using a copy made with D CONVERT/SHARE which was restored to the disk at the same time as theH bad version you need to replace with it, if the copy is kept on the sameI disk (which is a much quicker operation than pulling it off of an earlier  backup tape to replace it).   K CONVERT/SHARE on vulnerable files is a good plan. The only thing it doesn't M protect you from that I know of is the possibility of inconsistencies between = different files (such as SYSUAF and RIGHTSLIST, for example).    --- Carl   ------------------------------   Date: 5 Nov 2003 20:58 CST. From: carl@nospam.gerg.tamu.edu (Carl Perkins)C Subject: Re: Does anyone shut down for system disk backup any more? 3 Message-ID: <5NOV200320582416@nospam.gerg.tamu.edu>   % Z  <zarlenga@conan.ids.net> writes... & }Scott Vieth <svieth@wi.rr.com> wrote:E }: We've got a "procedure" in place where the computer operators shut F }: down our main VMS system once a month in the middle of the night toE }: backup the system disk.  They boot the Standalone Environment from 5 }: another disk and then backup the real system disk.  } + }: Does anyone still follow this procedure?  }.... G }: Should I just skip the "shutdown" and use the EVA to make a snapshot ; }: of the system disk?  Or just backup the system disk with  }: "/IGNORE=INTERLOCK"?  } H }I support a set of 365/24/7 transaction processing systems and I _only_H }make system backups after booting to CD.  I never back it up while it'sF }running with /ignore=interlock since there is a chance thatI will not }have a usable backup. }  }That would be catastrophic.  F But a lot less catastophic than having a zero percent chance of having= a good backup, which is what you have when you don't do them.   A Given a choice between doing a backup at time X that has a slight B chance of being bad and not doing a backup at time X at all, doing it is the better option.  ? Consider that you could keep your current "boot from CD" backup A schedule, but also add an additional "/ign=inter" backup half way A between each. This gives you an excellent chance of being able to > restore to a point less far back in the past than your current? schedule (especially if you use the CONVERT/SHARE thing to make  clean copies of various files).   ? The odds of there being file corruption in such a case are very A small. This means that there is a tiny chance that you might have D to swap a copy for one (or maybe more than one, but the odds of thatB are even smaller) of those files. The chances that that won't workH are really really small - and in such a cse you are where you would ahveD been if you hadn't tried it, but with some additional down time. TheB chances of this happeneing are probably not too different from theC chances of the tape your most recent "safe" backup is on going bad, 8 causing any restoration attempt to fail when you try it.  G No matter what you do, something can go wrong. The tape your supposedly G good "I booted from CD to get it" backup is on could fail, rendering it D useless. How catastrophic would that be? What do you do to deal withF the possibility? Is this better or worse than a small chance of one orD two corrupted files on a backup that is otherwise good (allowing youC to boot, but not function in the normal way from a user standpoint)hE and which has a good chance of being recoverable by renaming a one orlH two of the copied files and restarting the relevant things or rebooting?  @ Risk management is not about eliminating all risk for the simpleE reason that it is literally impossible to do so. You need to consider C your options and choose what to do based on the cost of the options F (in various areas - cost to implement it up front and continuing, timeE it takes from other things on an ongoing basis, and the time it takesoA to recover when things go wrong) compared with an estimate of them< probability of the events relevant to each option happening.  E As an example: Odds of building burning down: low (if this is not theyA case, you need to move to a different part of the planet).Odds of F both buildings housing the two parts of your disaster tolerant clusterH buring down even though they are 300 miles apart: even lower - but still@ possible. The first case can be dealt with by having the clusterB mentioned in the second case. Most companies don't do much to dealB with the second possibility as it is usually just too expensive toH deal with in a way that would keep you running more or less continuouslyD if the event happened compared with the probability of it happening.  @ So, in a way that is relevant to this discussion, what do I mean> by this? While the backup on a running system disk may produceF unrecoverable files, this risk can be mitigated by using CONVERT/SHAREG on vulnerable files - this operation has a very low cost (a few seconds B of time and maybe a couple millimeters of tape). It can be furtherD mitigated by doing the occasional backup by booting from CD. You mayB not be able to afford to shut down once a month (or however often B you'd prefer) to get a clean backup of your system disk - but manyB cases where that is true may be able to shut down once a year, and> do backups while the system is active (taking advantage of the= CONVERT/SHARE operations to reduce the risk) at more frequentR' intervals between those annual backups.-  = A backup that has a very small chance of being bad is roughly-< infinately better than not having a backup from that time: a9 nearly 100% chance of having a good backup compared to anb exactly 0% chance.  A Clearly if you can not shut the system down to get a clean backupnA as often as you think you should have your system disk backed up,sB your one and only usefull option is to do something that gives you@ backups while the system is up in conjuntion with doing anythingE that it is practical to do to increase the probability of the backups A being good. A good example would be to shut off anything you can, ? do a CONVERT/SHARE on the files that this can help recover, and @ then do the backup with /IGNORE=INTERLOCK and hope for the best.C In such a case there is hope - but there is none if you do nothing.o   --- Carl   ------------------------------  % Date: Wed, 05 Nov 2003 21:17:04 -0600 ( From: brandon@dalsemi.com (John Brandon)C Subject: Re: Does anyone shut down for system disk backup any more?.1 Message-ID: <03110521170408@dscis6-0.dalsemi.com>   9 > }That would eliminate the re-create as mentioned above.CP > }My policy - if I can define a logical and subsequently move a file off of the > }system disk - I am so there!, > D > I do not beleive that this solves the problem, it just moves it to > another disk.X  K It solves the problem of getting it off the system disk and helps eliminate J open files on the system disk - the discussion that has been at hand.  TheJ thread I responded to maintained that the queue database was on the system) disk.  I saw move it off to another disk.p  O Carl -  However you are correct that if the file is corrupt - regardles of what > disk it is on - it is still corrupt and will need to be fixed.       J*o*h*n B*r*a*n*d*o*nn VMS Systems Administratort* firstname.lastname.spam.me.not@dalsemi.com   ------------------------------   Date: 5 Nov 2003 21:22 CST. From: carl@nospam.gerg.tamu.edu (Carl Perkins)C Subject: Re: Does anyone shut down for system disk backup any more?03 Message-ID: <5NOV200321225674@nospam.gerg.tamu.edu>y  , brandon@dalsemi.com (John Brandon) writes...H }> An interesting experiment would be to find an old disk with an easilyM }> accessible "write protect" button on the outside.  Boot up VMS from such atJ }> disk, get your applications going (yes, even if they are on a differentE }> disk), and then write protect your system disk.  I wonder how manyo5 }> components would start failing and/or complaining?, } O }I sold that equipment years ago.  If memory serves me correctly the OS did notsP }like it at all and system activity was suspended... but I have slept since then }and can not really be certain.t }J*o*h*n B*r*a*n*d*o*n  J If your page and swap files are on that disk at the very least any processJ that needs them would hang. Loging in might be difficult as the last loginI time wouldn't be able to be written to the SYSUAF file (I don't know whatdK the login proceedure would do in such a case, it may still let you log in).   D At the very least, if the audit server discovers that it can't writeC to its log file (which will be the case if it wasn't moved from themB default location) I would expect it to lock the system down pretty& tightly, suspending nearly everything.   --- Carl   ------------------------------  # Date: Thu, 06 Nov 2003 05:36:13 GMT6% From: "Mike Naime" <mnaime@kc.rr.com> ! Subject: Re: Erp running on Vms ? 6 Message-ID: <h5lqb.1331$0u.1152@twister.rdc-kc.rr.com>  I I didn't know that Oracle left VMS.  We have Oracle Parallel Server (OPS)e running our App on VMS.e  G Can you explain what you mean by ERP?  I do not understand the Acronym.-  - labadie <labadie_g@decus.fr> wrote in message , news:bo93c8$9bk$1@news-reader3.wanadoo.fr...I > I know many customers are leaving Vms because they are moving to an Erp3+ > (Sap, Baan, Bpcs, Oracle applications...)  >o > Sap used to run on Vms.t > > > I wonder if there was an Erp available, today or in project. > D > I think it may be critical for the future of Vms (on Vax, Alpha or	 Itanium).d >r	 > Regardst >  > Grard >/   ------------------------------  % Date: Thu, 06 Nov 2003 04:39:49 -0000o7 From: Timothy Kupp <tkupp@entlinux.no.sasktel.spam.net>r, Subject: Expected transfer speeds using DSSI/ Message-ID: <vqjk4lpi1hk9a7@corp.supernews.com>t   VAX 7000	 VMS 5.5-2I  % Hardware	Advertised speed	DescriptionP% --------	----------------	-----------u7 KFSMA		4MB/s per port		DSSI adapter card with two ports ( HSD10		3MB/s			cabinet for SCSI devices; 					  connects to DSSI bush. TZ88		1.5MB/s uncompressed	SCSI DLT tape drive 		2.5MB/s compressed# RZ29?		9MB/s?			4GB SCSI hard drive   @ In our VAX 7000 we have 4 SCSI disks each installed in two HSD10G cabinets mounted inside the VAX.  We also have an external HSD10 with ahH TZ88 tape drive.  One internal HSD10 is connected to one port of a KFSMAD and the external HSD10 is chained to the other internal HSD10 to the other DSSI port.  E I'd like to hear from anyone else out there with a similar setup whatoC your transfer rates are.  Disk (in one HSD10) to disk (in the other-H HSD10) using the COPY command to copy a 500MB file, we are getting aboutG 1.2MB/s.  Disk to tape with the BACKUP command, we are peaking at aboutgD 1.5MB/s.  I figure (hope) we should be getting closer to the maximumF advertised transfer speed of the HSD10 at 3MB/s when copying from diskF to disk and to the maximum transfer speed of the tape drive at 2.5MB/s@ when backing up to it.  Wishful thinking?  Advertised speeds too	 generous?S  G One thought that occurred to me: I'm wondering if we're somehow hittingeH the 1000 I/Os per second limit of the HSD10 cabinet.  How do you monitor the I/O rate of an HSD10?   # Anyone have any suggestions/advice?d   ------------------------------  % Date: Wed, 05 Nov 2003 23:50:08 -0500 % From: Howard Shubs <howard@shubs.net>e% Subject: Re: Fortran calling sequence < Message-ID: <howard-E21AFC.23500805112003@enews.newsguy.com>  8 In article <8sbfqvkevajrsh1qruc1ds74r6k1js097u@4ax.com>,/  John Laird <nospam@laird-towers.org.uk> wrote:s  L > No, the argument block consists of a longword 1, followed by what is shown > as a .CKARG <address>.  F Fair enough.  I guess I could try firing up my VAX and feeding it the  code, but why bother?  :->  K > It would probably be clearer with a slightly more complicated example, ini= > fact.  Where there were other psect items before and after.    Yes.   -- oD You are what you eat, therefore, I'm a vegetable!  Cows and chickens and Pop Tarts are too.   ------------------------------  # Date: Wed, 05 Nov 2003 20:31:57 GMT * From: Paul Anderson <paul.anderson@hp.com>" Subject: Re: Full Duplex NIC cards5 Message-ID: <051120031530061743%paul.anderson@hp.com>a  A Isn't saying "NIC card" sort of like saying "ATM machine" or "PIN  number"?   Paul   -- 2  Paul Anderson   OpenVMS Engineerings   Hewlett-Packard Companye   ------------------------------   Date: 5 Nov 2003 22:43:01 -0800L. From: al5vf03p02@sneakemail.com (William Webb)" Subject: Re: Full Duplex NIC cards= Message-ID: <d5ce4b06.0311052243.1f8dd0a7@posting.google.com>u  g Paul Anderson <paul.anderson@hp.com> wrote in message news:<051120031530061743%paul.anderson@hp.com>...iC > Isn't saying "NIC card" sort of like saying "ATM machine" or "PINt
 > number"? >  > Paul     It's quite simple, actually.  D If we called them "NI" cards, we'd receive "shrubberies" by mistake.   WWWebb   ------------------------------  $ Date: Wed, 5 Nov 2003 20:16:40 -0500# From: "Hal Kuff" <kuff@comcast.net>t# Subject: Re: Help W/NAS proposal...c- Message-ID: <boc7d0$3ke@library2.airnews.net>L  0     Thanxs for the reply... here are my comments  0 "Mike Naime" <mnaime@kc.rr.com> wrote in message0 news:zc0qb.2020$j91.751@twister.rdc-kc.rr.com... > Hal: >EJ > First of all let me say that you seem to be approaching the DRM solution in > a bass-ackwards order. > K > There are basically 3 levels to transfer your data at.  I think that thisu( > lists them in the preferred order too. >i# > 1.) The Storage Controller level.i > 2.) The VMS OS level.  > 3.) The Network level. >7L > You have selected option #3  Potentially the slowest level to attempt thisJ > at.  Hopefully you are at least using GIG-E for network connectivity, or are  > you using 100MB network?    I     As stated, the bandwidth we have to work with is a t-3 that runs 2500y miles Maryland to Reno   > I > If you really have 5 LARGE StorageWorks SANS, do you have a maintenanceyK > contract on this equipment?  Who is your Storage TAM?  Get them to designnL > this for you.  If your local folks don't know anything, and your TAM isn'tI > much help, then get an Engineering rep pulled into the deal.  It soundst likeK > you are trying to design this on your own which makes me suspect that you  doL > not have a support contract, or good sales folks.  I would just ask one of7 > my TAMs these kinds of questions as a starting point.8 >$J > Note:  I define BIG at 25 TerraBytes (What I have in one EVA), and LARGE atH > 100 Terrabytes!  OR, 16 SAN switches in a single fabric.  What is your
 > definition?@ >iI > Who makes your SAN switches?  Are they Compaq branded Brocade switches?p IfE > so, Have you talked to Brocade about your Problem?  I was just at aI Brocade I > Users Group meeting tonight.  Among other topics, They where discussingoH > ISCSI and their newer/future product releases.  They would probably be able* > to help with a SAN-2-SAN based solution. >   $     We have Compaq 2gb SAN switches.  L > How much data do you need to replicate on a daily basis?  Can your IP pipeL > handle that?  If so, you should be able to do it between 2 VMS systems andK > skip the NAS head BS.  Your main problem as I see it is that your company H > does not want to fork over the buck$ to get you the bandwidth that youL > really need for making this a viable VMS solution....  So, you are turningC > to Microshit products to try and hide the problem.  DRM or remote 	 shadowingfK > is going to solve this at an easier complexity level than trying to writea > your VMS data to a NAS head.  8     Less than a T-3 can pump... which is > 20gb per hout >sJ > What happens to your VMS systems when the NAS head goes down?  Will this) > trash your data?  Have you tested this?mI > How much will sending this data to a NAS head slow down your systems???o >d >iL     Nothing... the NAS heads will wait for the link... we can tolerate shortG outage as the VMS system would not stall as it only sees the local diskn   >eL > If you are looking to make a Windows solution for a VMS storage pool, thenI > in my opinion, you are asking for major problems.  CASA, not NAS is the,K > remote linking solution that you need to investigate if you really insistn onL > going down the Microshit low bandwidth IP connectivity solution.  CASA was& > designed for remote IP connectivity.    L         We're n ot looking for a Windows solution... we're looking for an IPG link that runs at T-3 speeds over a 2500 mile link.... what do you knowe: about CASA because here is what the CASA announcement says  J StorageWorks arrays as well as those from EMC, Hitachi and Dell. SupportedK operating systems include HP-UX, Microsoft® Windows® NT and Windows 2000,-H Sun Solaris, Linux, IBM AIX and Novell NetWare. HP OpenView CASA 5.5U isH expected be available in May. More information about HP's virtualization. offerings and HP OpenView CASA is available at( http://www.hp.com/storage/virtualization  &     See anything about VMS.... I don't   >a >  >, > . > Hal Kuff <kuff@comcast.net> wrote in message) > news:bo9g1d$dq9@library2.airnews.net... 9 > > On Mon, 3 Nov 2003 11:11:41 -0500, Hoff Hoffman wroteF8 > > (in message <17vpb.8262$0b3.7098@news.cpqcorp.net>): > >n > >fH > >    In this case we are looking at RMS Journaling.... we need the AIJ filesr1 > > available to be applied to the remote system.e > >DE > > Well.... we have 5 large storageworks sans and have looked at the L > > continuous availability product but it is severely ill-defined and there > isL > > no supported way to connect the two systems over IP (the cisco 9216 with > IPJ > > module is not well understood at HP storage even though they sell it). > >y > > L > >    There are a lot of smart people here... it would appear that there is ahE > > way to take two NAS units, set them up to mirror a volume betweenh > BaltimoreiF > > and Reno and then have an OpenVMS system place AIJ files there forJ > > replication to the remote site.  The NAS boxes will handle everything, therJ > > question at hand is how to make the local NAS disk a target of VMS I/O > > operations...? > >e > >t > >  > >p > >i > >n > >e > >nA > > > In article <6MCdnVqhu5E0gDmiRVn-vQ@comcast.com>, "Hal Kuff"  > > > <kuff@tessco.com> writes:R > > >>K > > >> What we would like to do is mirror two disks across a T-3 ..... as as > T-3rI > > >> will pull about 20gb/hour of uncompressed data it seems reasonablei that > a J > > >> pair of HP NAS units could keep up with 10gb per hour..... We would > setupm > > arH > > >> pair of the NAS boxes on each end of the T-3 and then NFS mount a > volumeG > > >> from the NAS unit on an OpenVMS system at each end.  One OpenVMSn systemE > > >> would write ot the NFS disk constantly and the blocks would bel synched  > up > > >> with the other system.D > > >> > > >> Is anyone doing this? > > >.L > > >   OpenVMS host-based Volume Shadowing and OpenVMS Cluster software canG > > >   and does operate across a T3, assuming the distance involved is  withinC > > >   the SPD-specified limit.  This configuration would be fullyi
 supported. > > >eJ > > >   That written, OpenVMS host-based Volume Shadowing does not operateJ > > >   across the NFS protocol; it does not support data storage on disks > > >   that are NFS-served. > > >wL > > >   There are also a DRS-based (hardware-based) approaches that can alsoF > > >   remote mirroring (hardware folks often use the term mirroring, OpenVMSfH > > >   tends to use the term shadowing) of data.  You may or may not be able? > > >   to access the data "hot" on both ends -- that said, thes hardware-based8 > > >   solutions I've seen have a primary/active and an off-line/hot-standbyB > > >   design.  (OpenVMS Volume Shadowing allows distributed/peer
 read-write > > >   access.) > > >gJ > > >   If you want to roll your own transactional and shadowing protocol, > well,sJ > > >   have at.  (I'll assume you are aware of the volume of work you areH > > >   clearly inviting here, as this is not a trivial effort -- normalC > > >   operations are typically easy to code and operate, it's theb	 detectionmI > > >   and recovery from errors and failures that's a far more difficultp part > > >   of the work involved.) > > >-I > > >   I'd expect that an NFS-based scheme might well need to verify the< I/OeI > > >   using a read-after-write, and the simplest implementation of thisc willK > > >   significantly reduce bandwidth.  (MD5- or checksum-based scheme for K > > >   verifying the results -- with the assistance of computes and I/O onsL > > >   the remote server -- would obviously and potentially reduce the load > > >   on the wire itself.) > > >tG > > >   There are "lesser" approaches available as well, which snapshot. filesfJ > > >   and/or transactions across spindles and across a network, and that canrB > > >   allow the database server(s) to roll forward from the last checkpoint.hJ > > >   This isn't as nice as mirroring or shadowing, but it can be easier andvD > > >   cheaper -- and it trades off the ability to hot-start during	 failures;eA > > >   some resynchronization effort and/or some restart-focused  > site-specificdG > > >   procedures are involved as part of a failure-restart operation.T > > >  > > >s6 > > >  ---------------------------- #include <rtfaq.h># > > > -----------------------------t7 > > >     For additional, please see the OpenVMS FAQ --o > > www.hp.com/go/openvms/faq 8 > > >  --------------------------- pure personal opinion! > > > ---------------------------hK > > >         Hoff (Stephen) Hoffman   OpenVMS Engineering   hoff[at]hp.comd > > >  > >g > >  > >e >o >v   ------------------------------  $ Date: Wed, 5 Nov 2003 20:22:45 -0500# From: "Hal Kuff" <kuff@comcast.net>y# Subject: Re: Help W/NAS proposal...l- Message-ID: <boc7ob$9ms@library2.airnews.net>-  L     Hi,  thanxs for the reply... we really are just trying to replicate someH data between two sites over a t-3 at long distances... 2500 miles....  IJ don't see where HP has any intention of doing CASA with OpenVMS support...D can't seem to find anything... we do have HP support... 25,000/monthE maintenance bill... but honestly if you think for one moment that themI knowledge of these products is easy to come by...... It would be great ifrK som eone could point me to a real product that is avialble now and mentionsiA OpenVMS... CASA does not http://www.hp.com/storage/virtualizationi    0 "Hoff Hoffman" <hoff@hp.nospam> wrote in message, news:gcYpb.8419$w65.1614@news.cpqcorp.net...: > In article <bo9g1d$dq9@library2.airnews.net>, "Hal Kuff" <kuff@comcast.net> writes:8 > :On Mon, 3 Nov 2003 11:11:41 -0500, Hoff Hoffman wrote7 > :(in message <17vpb.8262$0b3.7098@news.cpqcorp.net>):n >aG >   You are seeking to roll your own mirroring, and this is not an easy E >   problem.  (It only looks like an easy problem.  The same could belH >   said about disaster tolerance -- while it might initially look easy,H >   there are just so many things that can go wrong that the problem canI >   and often does become quite difficult.  Again, not so much because ofSG >   the technology involved, but because of the vast number of failurest >   that can arise.) >eG > :   In this case we are looking at RMS Journaling.... we need the AIJl filesn0 > :available to be applied to the remote system. > K >   I don't know if it is possible to grab a hot copy of the RMS journalinga7 >   files.  That is something you will want to confirm.  >eD > :Well.... we have 5 large storageworks sans and have looked at theK > :continuous availability product but it is severely ill-defined and theree isK > :no supported way to connect the two systems over IP (the cisco 9216 withm IPI > :module is not well understood at HP storage even though they sell it).  > H >   I am not familiar with a "continuous availability product", at least  >   not by that particular name. > G >   The target here would be clustering, which -- over switches, thoughlK >   not over IP -- will do what you want, with host-based volume shadowing.o >oK > :   There are a lot of smart people here... it would appear that there is  aeD > :way to take two NAS units, set them up to mirror a volume between	 Baltimore-E > :and Reno and then have an OpenVMS system place AIJ files there forrI > :replication to the remote site.  The NAS boxes will handle everything,o therI > :question at hand is how to make the local NAS disk a target of VMS I/Om > :operations...?, >tG >   OpenVMS shadowing does not operate over IP protocols, though it cane# >   and does operate over T3 links.o >dI >   Controller-level operations tend to have difficulty ensuring multipleBI >   copies remain in synchronization across hosts -- this is part of what J >   the cluster connection manager and the quorum scheme will maintain forH >   you -- and particularly that the possibility of data partitioning isA >   avoided.  To maintain shared-write access, you need to ensurelG >   coordination and synchronization.    Put another way, a distributed G >   shared-write request involving an IP network link effectively seeks D >   to implement a connection manager and quorum scheme (or a direct >   analog) over the IP link.  > G >   There are controllers that provide a single writer and fail-over onCH >   occurances of errors -- in a way, this is akin to clustering back inH >   (as it was then known) VAX/VMS V3 vintage CI configurations.  DRM isG >   an example.  There are also the "last write wins" disk I/O schemes,.D >   such as that provided by NFS.  But I know of no synchronized andF >   distributed shared-write scheme available within storage hardware. >f >e >p( >  ---------------------------- #include' <rtfaq.h> -----------------------------i4 >     For additional, please see the OpenVMS FAQ --  www.hp.com/go/openvms/faq , >  --------------------------- pure personal# opinion ---------------------------nG >         Hoff (Stephen) Hoffman   OpenVMS Engineering   hoff[at]hp.coma >e   ------------------------------  % Date: Wed, 05 Nov 2003 20:56:42 +0100L2 From: martin@radiogaga.harz.de (Martin Vorlaender)J Subject: Re: In CSWS (Apache) CGI, what's the IP address of the client (br; Message-ID: <3fa955fa.524144494f47414741@radiogaga.harz.de>r   Dirk Munk (munk@home.nl) wrote:  > Barratt, Chris (FMC) wrote:eN > > There is an environment variable (symbol) called REMOTE_ADDR which has the7 > > IP of the client as it's value...check out the fileo* > > apache$root:[cgi-bin]test-cgi-vms.com. >CI > Is this indeed always the address of the browser, or can it also be theu7 > address of the proxy server that the client is using?-  H It's the IP address of the system requesting the resource. If the client, uses a proxy, then it's the proxy's address.  H > It seems the default behaviour of Apache (don't know about CSWS) is toL > show the  proxy server address, if (!) the client is using a proxy server.  > It's not an APache (or CSWS) thing, that's just the way it is.  J > I am trying to use build / use the limitipconn module for Apache, and inL > the  description of that module there is an example of how to patch ApacheF > (a patchy -: ) ) to show the IP address of the browser in all cases.  E This relies (like all methods to aquire the client's IP address) on amA header line the proxy *might* add. IIRC, some proxy software usestD HTTP-proxy-via (which comes out in a CGI program as HTTP_PROXY_VIA).   cu,a   Martin --  F                           | Martin Vorlaender  |  VMS & WNT programmer3  Cetero censeo            | work: mv@pdv-systeme.demF  Redmondem delendam esse. |   http://www.pdv-systeme.de/users/martinv/:                           | home: martin@radiogaga.harz.de   ------------------------------  % Date: Wed, 05 Nov 2003 13:47:15 -0700u" From: Dave Pitts <dpitts@cozx.com>7 Subject: Looking for a disk dump over TCP/IP connection., Message-ID: <k-6dnV3q6JFv_DSiRVn-hg@csd.net>   Hello:  D I've got a MicroVAX-II system with a disk that starting to make evil noises.a  B I'd like to dump the disk over a TCP connection (RDUMP ?) and bootD the image under the SIMH VAX emulator. If anybody has such a programD it would save me the time to write my own. The system is running the TGV multinet TCP stack.i   Thanks.  -- oM Dave Pitts                   PULLMAN: Travel and sleep in safety and comfort. F dpitts@cozx.com              My other RV IS a Pullman (Colorado Pine). http://www.cozx.com/~dpitts    ------------------------------  % Date: Wed, 05 Nov 2003 15:25:46 -0500e* From: JF Mezei <jfmezei.spamnot@istop.com>* Subject: Re: newbie problem installing VMS) Message-ID: <3FA95CC9.C6D71F36@istop.com>    William Webb wrote:r' > bridge, your DKA200 is a 104MB drive.  > G > IMHO, any attempt to do an /IMAGE restore is destined to fail because-H > any reasonably recent version of VMS is going to expect, nay, require, > a much larger restore target.   N The B saveset is about 50 megs if I remember correctly. So this would not be aI problem at that stage of the install.  After that, you'd have to manuallynM restore just the files you want from the other savesets. B contains enough toiK boot. Also, if you want to boot from B alone, you need to go to SYSBOOT andb change the startup procedure.d  I B saveset generates a system that boots from something different from theeC normal startup.com (which is essentially a procedure that continuesiL installation of VMS). If you change it to the normal file, you can then bootM and access a very limited set fo DCL commands and then manually restore thoseoM executables you need from the C saveset. (for instance, I don't recall if TPUe( comes with the B or C or D etc savesets.   ------------------------------  % Date: Wed, 05 Nov 2003 21:45:33 +0100r- From: WiseLYNX <wiselynx.naima@fastwebnet.it>V* Subject: Re: newbie problem installing VMS8 Message-ID: <lkdqb.21582$gA.12944@tornado.fastwebnet.it>  E > At the risk of sounding like the kid who suggests that you let somewC > air out of the tires to dislodge the truck that's stuck under theb' > bridge, your DKA200 is a 104MB drive.t > G > IMHO, any attempt to do an /IMAGE restore is destined to fail becauserH > any reasonably recent version of VMS is going to expect, nay, require, > a much larger restore target.f [snip]E > Start at about page 41 for the tables that discuss what hardware isL > supported. > RZ23's aren't.    H well, I did have a look to the docs you said (thanks, other interesting @ material :) and, if I correctly interpret the table about space H requirement for the basic installation of OpenVMS 7.3 for VAX, I should  need al least 32MB.d  K Anyway, as someone else pointed out, the problem I'm having is far before..-  ) Thanks anyway, expecially for the docs :)m   -- n    Wise	   O  LYNXr  8 GPG key: http://wiselynx.interfree.it/wiselynx.naima.ascA MD5 checksum: http://wiselynx.interfree.it/wiselynx.naima.asc.md5u   -----BEGIN GEEK CODE BLOCK-----  Version: 3.1I GE>$/IT/S d+ s: !a>? C++(+++)>$ U P>+ L++>++++ E--- W++ N ?o ?K w-- ?O ?MaG ?V>+ PS+ !PE Y+ PGP+ t- ?5 X- R+(*) tv- b+(++)>+++ DI(++) !D G e++ h!*@  r%>++ y-->+t ------END GEEK CODE BLOCK------    ------------------------------  % Date: Thu, 06 Nov 2003 04:43:39 -0000r7 From: Timothy Kupp <tkupp@entlinux.no.sasktel.spam.net>r3 Subject: Opinions on Nemonix Ultra SCSI XMI adaptere/ Message-ID: <vqjkbrjnk6fg7a@corp.supernews.com>    VAX 7000	 VMS 5.5-2o  H We are looking at installing a Nemonix Ultra SCSI XMI adapter in our VAXF and attaching some devices to it such as a DLT tape drive and some/allF of our SCSI disks via BA350 cabinets that we already have or, perhaps,  an HSG70 that we would purchase.  @ We currently have our SCSI devices attached to the VAX via HSD10" cabinets and DSSI cables/adapters.  F Anyone have any experience good or bad with this card?  It has (had) aG Compaq/HP part number (DS-KZMCA-AB), but HP seems to be rather skittish ( about supporting it, at least in Canada.  G Anyone manage to get a reasonable support contract in place from HP forlB this card?  HP is talking about charging CDN$600-700 per month forE replacement-only support if the card should fail.  The card itself isp US$5000 (new).  4 Anybody know a place that sells refurbished/cheaper?   ------------------------------  * Date: Wed, 5 Nov 2003 18:30:10 +0000 (UTC) From: david20@alpha2.mdx.ac.uk9 Subject: Re: Piping X-Windows traffic over ssh on OpenVMSt) Message-ID: <bobfji$mqi$1@news.mdx.ac.uk>t  p In article <de82e5e4.0311050614.1b1dfa9@posting.google.com>, joe.lofft@itec.mail.suny.edu (Joseph Lofft) writes:w >koehler@eisner.nospam.encompasserve.org (Bob Koehler) wrote in message news:<U8F7Tu+JbYP$@eisner.encompasserve.org>...at >> In article <de82e5e4.0311041108.1e8523ea@posting.google.com>, joe.lofft@itec.mail.suny.edu (Joseph Lofft) writes:G >> > Does anyone have an example setup of how to pipe X-Windows trafficoD >> > over SSH on OpenVMS.  I have not been able to acomplish this on, >> > OpenVMS the way it can be done on UNIX. >> h) >>    Sure, if you tell us what you want.t >>  I >>    Are you trying to connect using VMS as the X server or client?  Thec0 >>    ssh server or client?  Which TCP/IP stack? >>  J >>    Or in plainer speak, what commands are you trying to do (or emulate) >>    on what platform?  >> nE >>    For a complicated example, I have a fellow SSH from a local VMS I >>    system to a remote VMS system, starting an X11 client which tunnelsnE >>    back to the local VMS system and then gets pushed onto his VXT.gF >>    That's non-trivial and probably more complex than what you need. >> nJ >>    A very common problem is a login.com that does a set display commandE >>    based on the needs of a non-SSH connection.  That wipes out the 7 >>    correct display which SSH provides automatically.i >fG >I need to run Oracle's Universal Installer program to install databaseuE >code.  I have no problem getting remote X connections loggin in witht >telnet and doing: >:; >$ set display/create/transport=tcpip/node=1.2.3.4/server=1s >iD >I can then run whatever X application I need.  However I have foundG >that when I am working with remote sites, the response time with theseiF >applications is deathly slow, due to network congestion.  What I haveB >foudn on UNIX is that when you pipe the X traffic through SSH the >response time is much better. >d8 >This is what I am attempting to do on VMS with no luck.  G Has the final version of the  TCPIP services SSH server and client beeneO released ? Last time I looked it didn't support any forwarding - this was goingtN to be supported with the final release (which was supposed to be released with$ the next version of TCPIP services).  ; As to tunneling speeding things up - I find that suprising.nO I'd expect it to slow things down since you have the overhead of encrypting andr decrypting the traffic.i  
 David Webb VMS and Unix team leader CCSS Middlesex University i   ------------------------------  $ Date: Wed, 5 Nov 2003 17:01:36 -0500 From: norm.raphael@metso.com- Subject: Questions on VERIFY at process startoQ Message-ID: <OFA145A71B.5DEE27DC-ON85256DD5.00772C5B-85256DD5.0078ACF3@metso.com>r  J I have a check at the start of SYLOGIN.COM that turns VERIFY off if F$MODE () is BATCH or NETWORK. A Does this mean that VERIFY is ON by default if the process is notd INTERACTIVE and OFF if it is?   K What happens with DETACH processes?  In particular, I have an app that runs I from SYSTARTUP_VMS.COM that generates a detached process and that processtI spaws subprocesses during its lifecycle.  Do the subprocesses inherit therI VERIFY characteristics of the parent detached process, absent anything in ( them that might control VERIFY behavior?  9 Is there a pointer to where I can read up on this easily?    ------------------------------  # Date: Wed, 05 Nov 2003 21:42:05 GMTo( From: David Harrold <DHarrold@wi.rr.com> Subject: Re: SAN Switch Issueb@ Message-ID: <76d97d3c2b600957fedbbbbfd62f25e6@news.teranews.com>  M On 5 Nov 2003 10:46:30 -0800, Jack.Trachtman@vmmc.org (Jack Trachtman) wrote:e  6 >We have an HP StorageWorks SAN using first generation7 >switches (Brocade 2800s at 1Gb).  We have two pairs ofh8 >switches, each pair interconnected with an ISL.  We are8 >about to purchase additional switches and would like to3 >get the new generation (Brocade 3800? at 2Gb).  My 2 >question is, how can I combine our 2800s with the3 >3800?  Can I break the present ISL and put the new ! >switch between the old switches?   K I wouldn't break the existing ISLs.  That probably would disrupt traffic ineN the fabric.  (assuming you may have a server on one switch talking to an array on another switch.)   M Rather, just add an ISL from the new switch to each of the existing switches.i   For example:   Current config:r  (     +-----------+    ISL   +-----------+(     |  SWITCH1  |----------|  SWITCH2  |(     +-----------+          +-----------+       New Config:   (     +-----------+    ISL   +-----------+(     |  SWITCH1  |----------|  SWITCH2  |(     +-----------+          +-----------+           \                   /w            \                 /             \               /s               \ ISL         / ISL               \           /a                \         /               +-----------+i               |  SWITCH3  |W               +-----------+s      5 >Any suggestion/experiences would be helpful. Thanks.   L And depending on the amount of traffic, you may need more than 1 ISL betweenJ each switch.  IIRC, HP stated to me that 3 alphas could saturate an ISL.     Hope that helps.   Dave Harrold    N ..............................................................................N David Harrold                              E-Mail: David_Harrold at aurora.orgI Sr. Software Systems Engineer              Phone:          (414) 647-6204 I                                            Pager:          (414) 941-4634fG Aurora Health Care                         Fax:          (414) 647-4999e 3031 W. Montana Street Milwaukee, WI 53215k   ------------------------------  % Date: Thu, 06 Nov 2003 09:16:40 +1100 / From: Malcolm Wade <Malcolm.Wade@bikerider.com>y Subject: RE: SAN Switch IssueaC Message-ID: <200311052216.hA5MGeM07317@mail022.syd.optusnet.com.au>u   Jack,t  6 >We have an HP StorageWorks SAN using first generation7 >switches (Brocade 2800s at 1Gb).  We have two pairs ofl8 >switches, each pair interconnected with an ISL.  We are8 >about to purchase additional switches and would like to3 >get the new generation (Brocade 3800? at 2Gb).  Myi2 >question is, how can I combine our 2800s with the3 >3800?  Can I break the present ISL and put the newE! >switch between the old switches?c5 >Any suggestion/experiences would be helpful. Thanks.n   Been there, done that.  M The basis answer is yes BUT you need to (1) ensure you setup your new switch -N domain id first and (2) ensure the value of PIDformat is the same.  New 2Gbit S switches ship with PIDformat set to 1, the older ones are default to 0.  They must bK be the same.  PIDformat is used to determine the format of the PD (funnily aV enough!) and if all your switches have less than 16 ports you're fine with a value of S 0.  PIDformat can affect other things (like AIX hosts I believe) where LUN mapping  L is based on the SAN PID.  Don't change it unless you know what you're doing.  N The configshow command will tell you what the current value is; the configure ! command will allow you to set it.s  O To get you up and running with your new switch I'd sugest you set PIDformat on aA the new switch to 0 if that's what your live switches are set to.   N If you use switch zoneing etc make sure there are no active configs and zones 2 (best to clear them completely) on the new switch.  O As long as you switch core parameters are the same, break the link between the oP two existing switches, then use your new cables (SC-LC) to interconnect the new R to the old.  Of course, when you do this any activity between the two switch will G stop so you need to understand your SAN traffic etc and shutdown/close D* applications etc depending on your config.  E The console will tell you of any issues when you do the interconnect.e  5 The existing config will be copied to the new switch.i   Malcolmp         Malcolm.Wade@bikerider.com   ------------------------------  # Date: Thu, 06 Nov 2003 03:20:27 GMTt% From: "Mike Naime" <mnaime@kc.rr.com>h Subject: Re: SAN Switch Issues4 Message-ID: <%5jqb.662$0u.237@twister.rdc-kc.rr.com>  B Malcolm did a very good job of explaining the process and hazards.  F Right now, your design has 1 "hop" maximum between any 2 ports in yourH fabric.  If you want to keep this, I would second David's recommendationL about adding an additional ISL from each of your current switches to the newI switch.  This makes a triangle of your fabric.  but you have only one hoph% betwen any two points in your fabric.w  L Now, If you have servers on your first switch (Domain 1) and your storage isJ on the second switch (Domain 2).  All of this storage traffic travels downC your ISL's.  So, if you are adding another switch inbetween the twosG switches, you are forcing all of this traffic to travel through the newo switch.   E If you are planning for future growth, you might consider a Core-EdgeyL design, or a "Storage Centric" design.  HP has a SAN design guide that couldH help you.  Brocade is also very helpful.  Get on their website and get aJ local rep to contact you to answer any of your questions.  Also, dependingK on your area, you might be able to join a Brocade users group.  (Free food!S/ and good presentations/demo's of new products.)h  I Note:  Brocade recommends that you upgrade to 2.6.1a and 3.1 for the bestdK compatibility between the 2 switches.  A Switch firmware upgrade would also 6 be a good time to change your PID mode to 1 for future expansion/compatibility.    : Malcolm Wade <Malcolm.Wade@bikerider.com> wrote in message= news:200311052216.hA5MGeM07317@mail022.syd.optusnet.com.au...r > Jack,  > 8 > >We have an HP StorageWorks SAN using first generation9 > >switches (Brocade 2800s at 1Gb).  We have two pairs of-: > >switches, each pair interconnected with an ISL.  We are: > >about to purchase additional switches and would like to5 > >get the new generation (Brocade 3800? at 2Gb).  My 4 > >question is, how can I combine our 2800s with the5 > >3800?  Can I break the present ISL and put the newt# > >switch between the old switches?-7 > >Any suggestion/experiences would be helpful. Thanks.  >t > Been there, done that. >rG > The basis answer is yes BUT you need to (1) ensure you setup your newS switchI > domain id first and (2) ensure the value of PIDformat is the same.  Newn 2GbitXI > switches ship with PIDformat set to 1, the older ones are default to 0.s	 They must.L > be the same.  PIDformat is used to determine the format of the PD (funnilyL > enough!) and if all your switches have less than 16 ports you're fine with
 a value ofL > 0.  PIDformat can affect other things (like AIX hosts I believe) where LUN mappingaG > is based on the SAN PID.  Don't change it unless you know what you'res doing. >tE > The configshow command will tell you what the current value is; theg	 configureM# > command will allow you to set it.m >dC > To get you up and running with your new switch I'd sugest you sett PIDformat onC > the new switch to 0 if that's what your live switches are set to.p >gI > If you use switch zoneing etc make sure there are no active configs and2 zones 4 > (best to clear them completely) on the new switch. >eL > As long as you switch core parameters are the same, break the link between theoI > two existing switches, then use your new cables (SC-LC) to interconnectu the new G > to the old.  Of course, when you do this any activity between the two- switch will-H > stop so you need to understand your SAN traffic etc and shutdown/close, > applications etc depending on your config. >gG > The console will tell you of any issues when you do the interconnect.  >n7 > The existing config will be copied to the new switch.J > 	 > Malcolma >a >- >e >o > Malcolm.Wade@bikerider.com   ------------------------------  % Date: Wed, 05 Nov 2003 15:19:36 -0500K* From: JF Mezei <jfmezei.spamnot@istop.com>6 Subject: Re: To foreign or verb ? That is the question) Message-ID: <3FA95B57.BF55FA1D@istop.com>c   Tom Wade wrote:tG > Because the last time I asked this question here, the answer was "youtN > shouldn't be doing this anyway".  I am delighted to hear it is possible, and& > I'll have a read of the PCSI manual.  L The problem with the PRODUCT limitations is that to go around them, you needL to use the subprocess where your own command procedures can be executed. OneL problem, unless this has changed in recent times, is that the output of suchN procedures was sent to NLA0: unless the user specifically invoked PRODUCT with the right qualifier. p  G And another problem is that the subprocess doesn't benefit from all the-D environment created for the PCSI installation script and vice versa.  N PCSI reminds me a bit of IBM mainframes where you have JCL for batch jobs, andK CLIST for interactive jobs. With VMSINSTAL, it was the same language insidew the install procedure.   ------------------------------  % Date: Wed, 05 Nov 2003 22:33:30 +0100o" From: Didier Morandi <no@spam.com>B Subject: VAX/VMS to Itanium Migration News, November issue is here2 Message-ID: <3fa96cb8$0$244$636a55ce@news.free.fr>  C      English: www.didiermorandi.com/vms/vaxvms2itanium_200311en.pdfNC       French: www.didiermorandi.com/vms/vaxvms2itanium_200311fr.pdf:0                          soon on OpenVMS.org too  P I do not know why the .doc has clickable URLs and pictures, but when distilling M it, the .pdf has only clickable URLs but no more clickable pictures. Need to . investigate.  I Comments *very* welcome in here or: firstnamelastname a t nerim d o t net  Enjoy.   D. --F didier morandi  ~ sarl au capital de 8 000 euros ~  Revendeur agr HPC      Expertise en environnement DIGITAL ~ Formation ~ ProgrammationeF Offshore ~ 5 av. A. Durand 31700 Blagnac France. Tl: 33(0)5 6131 6287D    SIRET 448 694 851 00016 RCS Toulouse http://www.didiermorandi.com   ------------------------------  # Date: Wed, 05 Nov 2003 19:29:31 GMTs# From: "John Smith" <a@nonymous.com>y3 Subject: Re: VMS in pole position in an HP WEB pageaG Message-ID: <vccqb.22228$YN6.8337@news02.bloor.is.net.cable.rogers.com>a   Tim Llewellyn wrote: > Larry Kilgallen wrote: >>D >> In article <3fa75b4b$0$241$636a55ce@news.free.fr>, Didier Morandi >> <no@spam.com> writes:( >>> Scroll to "Platforms and technology"B >>> http://h71028.www7.hp.com/enterprise/cache/4244-0-0-0-121.aspx >>* >> Using Netscape Navigator 4.75-20000815,C >> under "Platforms and technology" I see a black box big enough tow
 >> hold text.eG >> Looking at the source, I see VMS would have been there.  Perhaps thes >> flawd> >> is one of the 28 errors found by http://validator.w3.org/ . >pD > These, days, people either seem to laugh at me or shrug when I say% > "it doesn't work with netscape 4.7.  >, > :-( RIP www.    L Opera will work on the same machine as IE. I've just started trying it as an alternative to IE.   ------------------------------  % Date: Thu, 06 Nov 2003 02:03:07 +0800u, From: Paul Repacholi <prep@prep.synonet.com>- Subject: Re: VMS not mentioned by HP....againh- Message-ID: <87brrq3d7o.fsf@prep.synonet.com>u  Q Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> writes:d  @ > Itanium has been available since 2001, HP-UX on Itanium ditto.   > How long do they need ??  F Andrew, how could you. Just because it has missed the year, decade and? century of release is no reason to question its timelyness. And E remember, if you want integrity from HP, from now on you have to tello+ them is you want a rack mount kit for it :)      --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda.h@                                              West Australia 6076* comp.os.vms,- The Older, Grumpier Slashdot. Raw, Cooked or Well-done, it's all half baked.F EPIC, The Architecture of the future, always has been, always will be.   ------------------------------  % Date: Wed, 05 Nov 2003 21:29:08 +0000t) From: Antonio Carlini <arcarlini@iee.org>  Subject: Re: VUP?o? Message-ID: <EYdqb.3438$382.1712@newsfep4-winn.server.ntli.net>e   JF Mezei wrote:tO > Since the VUP was used as a MIPS style of measure, I would venture that the PaM > was for performance.  Model XX had 31 VUPS means that it performed 30 timesl > faster than a 780.  A No if it was 31 VUPs it was 31 times *faster* than a VAX-11/780, SG although I'll concede that it's performance was only 30 times *greater*e" (or 3000% greater, if you prefer).  N > And the S is simply the plurial of VUP in my opinion. The 780 had 1 VUP, the > 9000 had 2 VUPS., > (or whatever the 9000's rating truly was).  E 40 VUPs for one CPU, 79 for 2, 118 for 3 and 157 for 4. Only ever sawoE one once in real life ... I always wondered who had the other one :-)w   AntonioA   -- c   --   ---------------s- Antonio Carlini             arcarlini@iee.orgD   ------------------------------  * Date: Wed, 5 Nov 2003 16:32:12 -0600 (CST) From: sms@antinode.org Subject: Re: VUP?n) Message-ID: <03110516321247@antinode.org>   ) From: Antonio Carlini <arcarlini@iee.org>n   > JF Mezei wrote:aQ > > Since the VUP was used as a MIPS style of measure, I would venture that the PtO > > was for performance.  Model XX had 31 VUPS means that it performed 30 timesf > > faster than a 780. > C > No if it was 31 VUPs it was 31 times *faster* than a VAX-11/780, oI > although I'll concede that it's performance was only 30 times *greater*.$ > (or 3000% greater, if you prefer).  H    Well, duh.  If its "performance was only 30 times *greater*", then itH was only "30 times faster", not 31.  Like Mr. Mezei, I assume, it drivesC me to distraction when people can't distinguish between "N times astC much" and "N times more".  But then I can also use "its" and "it's" 
 correctly.  ? > > And the S is simply the plurial of VUP in my opinion. [...]i  D    Which is why it makes more sense to write VUPs, not VUPS.  (And IH also know that MIPS is not the plural of MIP.)  My recollection, too, is! that the P was for "performance".   F    Of course, in my opinion, pathetic English language skills would beE among the best reasons for excluding one fellow at Sun in the UK fromt, this news group, but that's just my opinion.  H ------------------------------------------------------------------------  4    Steven M. Schweda               (+1) 651-699-98183    382 South Warwick Street        sms@antinode-org     Saint Paul  MN  55105-25475   ------------------------------  $ Date: Wed, 5 Nov 2003 18:24:09 -0500 From: norm.raphael@metso.com Subject: Re: VUP? Q Message-ID: <OFB997419C.A83B1230-ON85256DD5.00800456-85256DD5.00803B88@metso.com>o  H We called that unit (VUP) the Vague Unit of Performance (nee Vax Unit ofJ Performance).  A MicroVAX was rated at 0.9 - 1.1 VUP's.  That's because of< differences IIRC in instruction emulation and Floating Point7 implementations.  So right away the Unit was imprecise.   J I prefer to add apostrophe "s" to make plurals of abbreviations, but there: is no fixed style rule for acronyms, pronounceable or not.I VUP is a pronounceable acronym.  RMS is an unpronounceable acronym.  SCSIn8 is an unpronounceable acronym that is pronounced anyway.  . From:  sms@antinode.org on 11/05/2003 05:32 PM  " Please respond to sms@antinode.org       To:    Info-VAX@Mvb.Saic.Com cc:c   Subject:    Re: VUP?    ) From: Antonio Carlini <arcarlini@iee.org>    > JF Mezei wrote:lK > > Since the VUP was used as a MIPS style of measure, I would venture thats the PdI > > was for performance.  Model XX had 31 VUPS means that it performed 30r timesa > > faster than a 780. > B > No if it was 31 VUPs it was 31 times *faster* than a VAX-11/780,I > although I'll concede that it's performance was only 30 times *greater*w$ > (or 3000% greater, if you prefer).  H    Well, duh.  If its "performance was only 30 times *greater*", then itH was only "30 times faster", not 31.  Like Mr. Mezei, I assume, it drivesC me to distraction when people can't distinguish between "N times asiC much" and "N times more".  But then I can also use "its" and "it's"w
 correctly.  ? > > And the S is simply the plurial of VUP in my opinion. [...]a  D    Which is why it makes more sense to write VUPs, not VUPS.  (And IH also know that MIPS is not the plural of MIP.)  My recollection, too, is! that the P was for "performance".c  F    Of course, in my opinion, pathetic English language skills would beE among the best reasons for excluding one fellow at Sun in the UK fromo, this news group, but that's just my opinion.  H ------------------------------------------------------------------------  4    Steven M. Schweda               (+1) 651-699-98183    382 South Warwick Street        sms@antinode-orgh    Saint Paul  MN  55105-2547    ------------------------------  # Date: Wed, 05 Nov 2003 23:47:10 GMT L From: winston@SSRL.SLAC.STANFORD.EDU ("Alan Winston - SSRL Admin Cmptg Mgr") Subject: Re: VUP?t6 Message-ID: <00A2873C.89ECBAFE@SSRL.SLAC.STANFORD.EDU>  k In article <EYdqb.3438$382.1712@newsfep4-winn.server.ntli.net>, Antonio Carlini <arcarlini@iee.org> writes:  >JF Mezei wrote:P >> Since the VUP was used as a MIPS style of measure, I would venture that the PN >> was for performance.  Model XX had 31 VUPS means that it performed 30 times >> faster than a 780.r >sB >No if it was 31 VUPs it was 31 times *faster* than a VAX-11/780, H >although I'll concede that it's performance was only 30 times *greater*# >(or 3000% greater, if you prefer).a >iO >> And the S is simply the plurial of VUP in my opinion. The 780 had 1 VUP, thec >> 9000 had 2 VUPS. - >> (or whatever the 9000's rating truly was).d >iF >40 VUPs for one CPU, 79 for 2, 118 for 3 and 157 for 4. Only ever sawF >one once in real life ... I always wondered who had the other one :-)  J SLAC had one.  Don't know if it was the one you saw.  I never actually sawD the one that was here, but I used to hear a lot of cursing about it.   -- Alan- --  O ===============================================================================j0  Alan Winston --- WINSTON@SSRL.SLAC.STANFORD.EDUM  Disclaimer: I speak only for myself, not SLAC or SSRL   Phone:  650/926-3056eM  Paper mail to: SSRL -- SLAC BIN 99, 2575 Sand Hill Rd, Menlo Park CA   94025nO ===============================================================================    ------------------------------   Date: 5 Nov 2003 21:48 CST. From: carl@nospam.gerg.tamu.edu (Carl Perkins) Subject: Re: VUP?h3 Message-ID: <5NOV200321480419@nospam.gerg.tamu.edu>u   arcarlini@iee.org writes...  }JF Mezei wrote:O }> And the S is simply the plurial of VUP in my opinion. The 780 had 1 VUP, thep }> 9000 had 2 VUPS.t- }> (or whatever the 9000's rating truly was).  } F }40 VUPs for one CPU, 79 for 2, 118 for 3 and 157 for 4. Only ever sawF }one once in real life ... I always wondered who had the other one :-) }  }Antonio  K There was one here at Texas A&M University. It even had a vector processor.   H A note that was posted the a VAX Notes forum that ran on that cluster at# the time (1991) said this about it:    [...]o/     ZEUS   VAX 9000-210 256 Mbytes    40.0 VUPSe [...]b  K     The VAX 9000's Vector processor has a peak vector rating of 125 MFLOPS.nJ     Compared to the CRAY Y-MP at 333 MFLOPS and the IBM 3090 at 80 MFLOPS.I     The VAX 9000 was acquired for about 1/8 the cost of the CRAY and even #     a smaller fraction of the 3090.i  H     With well vectorized code the 9000 can reach approximately 400 VUPS.   --- Carl   ------------------------------  # Date: Thu, 06 Nov 2003 00:45:08 GMT 4 From: brad@.gateway.2wire.net (Bradford J. Hamilton)' Subject: [OT] - VAX9000 (was: Re: VUP?)b/ Message-ID: <oQgqb.114837$Fm2.102912@attbi_s04>h   In article <00A2873C.89ECBAFE@SSRL.SLAC.STANFORD.EDU>, winston@SSRL.SLAC.STANFORD.EDU ("Alan Winston - SSRL Admin Cmptg Mgr") writes:al !In article <EYdqb.3438$382.1712@newsfep4-winn.server.ntli.net>, Antonio Carlini <arcarlini@iee.org> writes: !>JF Mezei wrote:2 !snip!P !>> And the S is simply the plurial of VUP in my opinion. The 780 had 1 VUP, the !>> 9000 had 2 VUPS.. !>> (or whatever the 9000's rating truly was). !>G !>40 VUPs for one CPU, 79 for 2, 118 for 3 and 157 for 4. Only ever sawuG !>one once in real life ... I always wondered who had the other one :-)  ! K !SLAC had one.  Don't know if it was the one you saw.  I never actually sawpE !the one that was here, but I used to hear a lot of cursing about it.S !i  E We had two (not clustered together).  One was a lemon, parts replacedhN constantly.  The other (hot standby) worked perfectly, when asked.  We finallyP smartened up towards the end, and made the "hot standby" the "primary" machine. G Too late, as it turned out - management forbade us from using them, and % negotiated VAX6000's as replacements.o   !snip!  J __________________________________________________________________________A Bradford J. Hamilton                    "All opinions are my own" K bMradAhamiPltSon-at-coMmcAast.nPeSt     "Lose the MAPS, and replace '-at-' h0                                          with @"   ------------------------------  $ Date: Wed, 5 Nov 2003 17:14:49 -0800# From: "Tom Linden" <tom@kednos.com>m+ Subject: RE: [OT] - VAX9000 (was: Re: VUP?)r9 Message-ID: <CIEJLCMNHNNDLLOOGNJICEKLIHAA.tom@kednos.com>y  " Was the 9000 the one based on ECL?   >-----Original Message-----n< >From: Bradford J. Hamilton [mailto:brad@.gateway.2wire.net]+ >Sent: Wednesday, November 05, 2003 4:45 PMr >To: Info-VAX@Mvb.Saic.Com( >Subject: [OT] - VAX9000 (was: Re: VUP?) >r >d7 >In article <00A2873C.89ECBAFE@SSRL.SLAC.STANFORD.EDU>,sA >winston@SSRL.SLAC.STANFORD.EDU ("Alan Winston - SSRL Admin CmptgV >Mgr") writes:A >!In article <EYdqb.3438$382.1712@newsfep4-winn.server.ntli.net>,t, >Antonio Carlini <arcarlini@iee.org> writes: >!>JF Mezei wrote: >!snip!fB >!>> And the S is simply the plurial of VUP in my opinion. The 780 >had 1 VUP, ther >!>> 9000 had 2 VUPS.-/ >!>> (or whatever the 9000's rating truly was).- >!>-H >!>40 VUPs for one CPU, 79 for 2, 118 for 3 and 157 for 4. Only ever sawH >!>one once in real life ... I always wondered who had the other one :-) >!L >!SLAC had one.  Don't know if it was the one you saw.  I never actually sawF >!the one that was here, but I used to hear a lot of cursing about it. >! > F >We had two (not clustered together).  One was a lemon, parts replacedC >constantly.  The other (hot standby) worked perfectly, when asked.s > We finally= >smartened up towards the end, and made the "hot standby" thee >"primary" machine.@H >Too late, as it turned out - management forbade us from using them, and& >negotiated VAX6000's as replacements. >i >!snip!e > K >__________________________________________________________________________bB >Bradford J. Hamilton                    "All opinions are my own"K >bMradAhamiPltSon-at-coMmcAast.nPeSt     "Lose the MAPS, and replace '-at-'n1 >                                         with @"e >s >---' >Incoming mail is certified Virus Free. ; >Checked by AVG anti-virus system (http://www.grisoft.com).vB >Version: 6.0.530 / Virus Database: 325 - Release Date: 10/22/2003 >a ---s& Outgoing mail is certified Virus Free.: Checked by AVG anti-virus system (http://www.grisoft.com).A Version: 6.0.530 / Virus Database: 325 - Release Date: 10/22/2003l   ------------------------------  % Date: Wed, 05 Nov 2003 21:54:50 +0100 - From: WiseLYNX <wiselynx.naima@fastwebnet.it>hH Subject: [semi-OT] SCSI jumpers (was: Re: newbie problem installing VMS)8 Message-ID: <4tdqb.21587$gA.11835@tornado.fastwebnet.it>  B > Getting back to his original problem, as the owner/operator of aD > number of VS3100's I have seen that error on many occaisions.  TheE > only solution I have found so far is to resort to a known supported F > CD Drive.  Mitsumi or Toshiba seem to come to mind.  OK, I just wentB > in the computer room and looked.  I am using Toshiba XM-4101B's.  A mhh.. In facts, I knew that VAXes are quite overnice about cdrom -H drives.. I saw that mine is a Matsushita (even if marked by IBM). Now I H have to find where is the list of well tested-and-working cdrom drives,  but should not be a problem.  H Anyway, going a bit OT, I'd be curious to know what do exactly mean the H various settings I can change with jumpers on the back of the drive. On G mine, beyond the three marked "1" "2" and "4", and are used to set the oE device ID, there are some other. One is pretty simple to understand,  E called "terminator". Then, there are three of which I miss a bit the -G meaning: one called "parity", one called "term power", and one "sector c size". Can anyone explain?  H (note: I did already try every combination of these also, and got still . no result.. it was just a curiosity of mine :)   -- 0    Wise	   O  LYNX0  8 GPG key: http://wiselynx.interfree.it/wiselynx.naima.ascA MD5 checksum: http://wiselynx.interfree.it/wiselynx.naima.asc.md5p   -----BEGIN GEEK CODE BLOCK-----r Version: 3.1I GE>$/IT/S d+ s: !a>? C++(+++)>$ U P>+ L++>++++ E--- W++ N ?o ?K w-- ?O ?MdG ?V>+ PS+ !PE Y+ PGP+ t- ?5 X- R+(*) tv- b+(++)>+++ DI(++) !D G e++ h!*@  r%>++ y-->+y ------END GEEK CODE BLOCK------r   ------------------------------  % Date: Wed, 05 Nov 2003 17:46:00 -0600i/ From: Chris Scheers <chris@applied-synergy.com>sL Subject: Re: [semi-OT] SCSI jumpers (was: Re: newbie problem installing VMS)3 Message-ID: <3FA98BB8.D4A6D1E9@applied-synergy.com>l   WiseLYNX wrote:b > D > > Getting back to his original problem, as the owner/operator of aF > > number of VS3100's I have seen that error on many occaisions.  TheG > > only solution I have found so far is to resort to a known supportedoH > > CD Drive.  Mitsumi or Toshiba seem to come to mind.  OK, I just wentD > > in the computer room and looked.  I am using Toshiba XM-4101B's. > B > mhh.. In facts, I knew that VAXes are quite overnice about cdromI > drives.. I saw that mine is a Matsushita (even if marked by IBM). Now I I > have to find where is the list of well tested-and-working cdrom drives,r > but should not be a problem. > I > Anyway, going a bit OT, I'd be curious to know what do exactly mean the5I > various settings I can change with jumpers on the back of the drive. OndH > mine, beyond the three marked "1" "2" and "4", and are used to set theF > device ID, there are some other. One is pretty simple to understand,F > called "terminator". Then, there are three of which I miss a bit theH > meaning: one called "parity", one called "term power", and one "sector > size". Can anyone explain?  H The problem with these jumpers is that some drives want the jumper on to& enable the function, some want it off.  H "Parity" indicates whether or not parity checking is done on data passed* over the SCSI bus.  I always turn this on.  F "Term power" indicates whether or not termination power is provided toE the SCSI bus.  There should be one device on the bus that does this. eB Generally, it is the host adapter card.  I normally turn this off.  H "Sector size" should, strangely enough, set the sector size.  If this is* a CDROM, this should be a 512/2048 jumper.    G ----------------------------------------------------------------------- $ Chris Scheers, Applied Synergy, Inc.  C Voice: 817-237-3360            Internet: chris@applied-synergy.com o   Fax: 817-237-3074P   ------------------------------   End of INFO-VAX 2003.615 ************************