1 INFO-VAX	Thu, 20 Apr 2006	Volume 2006 : Issue 219       Contents:5 Re: Error codes from Distributed NetBeans for OpenVMS  Re: Free rz26s and rz28s Re: Free rz26s and rz28s Re: Free rz26s and rz28s Re: Free rz26s and rz28s Re: Free rz26s and rz28s Re: Free rz26s and rz28s Re: Free rz26s and rz28s, Re: How does a fix become a published patch?, Re: How does a fix become a published patch?7 INK JET concentrate Manufacturer ariel@refillkit.com.ar E RE: Is OpenVMS certified yet for 64-way Hard partitions (GS1280 M64).  Re: OT: Sparc not dead yet< Re: Partitioned cluster question (reboot during lost quorum)< Re: Partitioned cluster question (reboot during lost quorum)< Re: Partitioned cluster question (reboot during lost quorum)< Re: Partitioned cluster question (reboot during lost quorum)< Re: Partitioned cluster question (reboot during lost quorum)< Re: Partitioned cluster question (reboot during lost quorum)< Re: Partitioned cluster question (reboot during lost quorum)* Re: Quorum, locks and application question* Re: Quorum, locks and application question* Re: Quorum, locks and application question Recommended sources for cables" Re: Recommended sources for cables" Re: Recommended sources for cables Sending mail via TLS. Re: Strange behavior of DEFINE with equal-sign. Re: Strange behavior of DEFINE with equal-sign The Undocumented OpenVMS VAX hardware error code 3 ?  Re: VAX hardware error code 3 ? > Re: VAX Q-bus and SCSI Disks?  (was: Re: Free rz26s and rz28s)% Re: What (WAS) wrong with my network?   F ----------------------------------------------------------------------    Date: 20 Apr 2006 01:56:49 -0700 From: jazzfahrer@gmail.com> Subject: Re: Error codes from Distributed NetBeans for OpenVMSB Message-ID: <1145523409.371026.21040@e56g2000cwe.googlegroups.com>   Hello newsgroup   8 With the support of Meg Watson I could solve my problem.  G The problem was in fact very simple. I didn't use the actual version of F Distribute NetBeans for OpenVMS. My version was 1.1 and the new one is 1.1-1.  E After installing the new version everything runs perferctly. I am now : able to start remote compile and build for Cobol programs.  
 Best regards, 
 Jazzfahrer   ------------------------------  % Date: Thu, 20 Apr 2006 14:53:52 +0300 4 From: Mike Rechtman <michael.rechtman.nospam@hp.com>! Subject: Re: Free rz26s and rz28s & Message-ID: <4447A080.4B2442D9@hp.com>   Richard B. Gilbert wrote:  >  > JF Mezei wrote:  >  > > syslost wrote: > > I > >>I have 20 rz26s and 130 rz28s, that for only the cost of shipping can C > >>be yours.  These are for Hobbyist use only.  eMail me if you're  > >>interested.  > >  > > L > > You should provide your location so that an estimate on cost fo shippingB > > can be made, as well as provide the capacity for those drives. > J > Anybody who doesn't rembmer 1.02GB for RZ26 and 2.04GB for RZ28 probablyG > doesn't want one.  In their day, ten to fifteen years or so ago, they H > were nice little disks; when I worked for McGraw-Hill back in 97-98 weG > had at least 200 RZ26s.  Today, they are more nostalgia than storage!    Wadda ya mean "in their day"? 7 written on a running *working* VAXstation 4000-60 with: D Device         Type   Max     Free     %   Clu.   Name         Open 
 Trans  ErrH DANDAN$DKA100: RZ28    2.0 GB  0.1 GB   5     4   DANDAN$BACK     1      2    0G DANDAN$DKA200: RZ25    406 MB  214 MB  53     3   DANDAN$USER    27      30    0 F DANDAN$DKA300: RZ26   1001 MB  620 MB  62     3   DANDAN062     543    415    0, DANDAN$DKA500: RRD43  device is not mounted.   Mike --  E --------------------------------------------------------------------- E Usual disclaimer: All opinions are mine alone, perhaps not even that. ? Mike Rechtman                            *rechtman@tzora.co.il* F Kibbutz Tzor'a.                          Voice (home): 972-2-9908337  B   "20% of a job takes 80% of the time, the rest takes another 80%"E ---------------------------------------------------------------------  -----BEGIN GEEK CODE BLOCK-----  Version: 3.1: GCM/CS d(-)pu s:+>:- a++ C++ U-- L-- W++ N++ K? w--- V+++$6 PS+ PE-- t 5? X- tv-- b+ DI+ D-- G e++ h--- r+++ y+++@ ------END GEEK CODE BLOCK------    ------------------------------  # Date: Thu, 20 Apr 2006 14:39:05 GMT * From: Allisonnospam@nouce.bellatlantic.net! Subject: Re: Free rz26s and rz28s 8 Message-ID: <t47f42t1d348f21d9n2i1d07vke9kcpegm@4ax.com>  6 On Tue, 18 Apr 2006 10:57:59 -0700, foo@bar.com wrote:  A >On 18 Apr 2006 10:37:02 -0700, "syslost" <wm.reynolds@gmail.com>  >wrote:  > G >>I have 20 rz26s and 130 rz28s, that for only the cost of shipping can A >>be yours.  These are for Hobbyist use only.  eMail me if you're 
 >>interested. ' >Would RZ-26s work on a MVax 3110 10e?  + >------------------------------------------ D >NO, use John Mee3 at comcast dot net (remove the spaces and do the $ >obvious with the "dot" and the "at"   Yes!  E The Boot disk is limited to the RZ26 (1.07mb max size) but second and - third disks can be larger (but not bootable).   B My 3100s easily fit three drives if there is no Floppy/tk70/Crdom.@ However with two or more drives make sure the fans work and the  dust is cleaned out.   Allison    ------------------------------  % Date: Thu, 20 Apr 2006 08:21:39 -0700  From: foo@bar.com ! Subject: Re: Free rz26s and rz28s 8 Message-ID: <ul9f429hjsclceoqgf1db1mj6cba6oaflo@4ax.com>  1 On Thu, 20 Apr 2006 14:53:52 +0300, Mike Rechtman ' <michael.rechtman.nospam@hp.com> wrote:    >Richard B. Gilbert wrote: >>   >> JF Mezei wrote: >>   >> > syslost wrote:  >> >J >> >>I have 20 rz26s and 130 rz28s, that for only the cost of shipping canD >> >>be yours.  These are for Hobbyist use only.  eMail me if you're >> >>interested. >> > >> >M >> > You should provide your location so that an estimate on cost fo shipping C >> > can be made, as well as provide the capacity for those drives.  >>  K >> Anybody who doesn't rembmer 1.02GB for RZ26 and 2.04GB for RZ28 probably H >> doesn't want one.  In their day, ten to fifteen years or so ago, theyI >> were nice little disks; when I worked for McGraw-Hill back in 97-98 we H >> had at least 200 RZ26s.  Today, they are more nostalgia than storage! >  >Wadda ya mean "in their day"?8 >written on a running *working* VAXstation 4000-60 with:E >Device         Type   Max     Free     %   Clu.   Name         Open   >Trans  Err I >DANDAN$DKA100: RZ28    2.0 GB  0.1 GB   5     4   DANDAN$BACK     1       >2    0 H >DANDAN$DKA200: RZ25    406 MB  214 MB  53     3   DANDAN$USER    27     >30    0G >DANDAN$DKA300: RZ26   1001 MB  620 MB  62     3   DANDAN062     543    	 >415    0 - >DANDAN$DKA500: RRD43  device is not mounted.  >   B So anybody know where I can get  SCSI adapters for my Mvax IIs????  C Seriously, I have two Microvax IIs that fine, except for one of the - hard drives in the BA123 case is a bit dodgy. * ------------------------------------------C NO, use John Mee3 at comcast dot net (remove the spaces and do the  # obvious with the "dot" and the "at"    ------------------------------  % Date: Thu, 20 Apr 2006 16:34:04 +0100 * From: "Richard Brodie" <R.Brodie@rl.ac.uk>! Subject: Re: Free rz26s and rz28s 2 Message-ID: <e289lc$4gu$1@blackmamba.itd.rl.ac.uk>  Q <foo@bar.com> wrote in message news:ul9f429hjsclceoqgf1db1mj6cba6oaflo@4ax.com...   D > So anybody know where I can get  SCSI adapters for my Mvax IIs????  C Something like a CQD-220, any number of places, if you've got a few D hundred dollars to spare and need it to work. Otherwise it's Ebay or dumpster diving, probably.     ------------------------------  % Date: Thu, 20 Apr 2006 09:46:30 -0700  From: foo@bar.com ! Subject: Re: Free rz26s and rz28s 8 Message-ID: <99ef42pi6954pv7vsqii3eo98tbsq34ght@4ax.com>  4 On Thu, 20 Apr 2006 16:34:04 +0100, "Richard Brodie" <R.Brodie@rl.ac.uk> wrote:   > R ><foo@bar.com> wrote in message news:ul9f429hjsclceoqgf1db1mj6cba6oaflo@4ax.com... > E >> So anybody know where I can get  SCSI adapters for my Mvax IIs????  > D >Something like a CQD-220, any number of places, if you've got a fewE >hundred dollars to spare and need it to work. Otherwise it's Ebay or  >dumpster diving, probably.   = Does anybody remember what the DEC part number was for a SCSI E controller that would support both tape and disk?  Trying to cover my % bases for when the ESDI drives die... * ------------------------------------------C NO, use John Mee3 at comcast dot net (remove the spaces and do the  # obvious with the "dot" and the "at"    ------------------------------  # Date: Thu, 20 Apr 2006 17:22:37 GMT * From: Allisonnospam@nouce.bellatlantic.net! Subject: Re: Free rz26s and rz28s 8 Message-ID: <logf429dlh2qifjfmloqufv33eg5e18g47@4ax.com>  6 On Thu, 20 Apr 2006 08:19:33 -0700, foo@bar.com wrote:  G >On Thu, 20 Apr 2006 14:39:05 GMT, Allisonnospam@nouce.bellatlantic.net  >wrote:  > 8 >>On Tue, 18 Apr 2006 10:57:59 -0700, foo@bar.com wrote: >>C >>>On 18 Apr 2006 10:37:02 -0700, "syslost" <wm.reynolds@gmail.com> 	 >>>wrote:  >>> I >>>>I have 20 rz26s and 130 rz28s, that for only the cost of shipping can C >>>>be yours.  These are for Hobbyist use only.  eMail me if you're  >>>>interested. ) >>>Would RZ-26s work on a MVax 3110 10e?  - >>>------------------------------------------ F >>>NO, use John Mee3 at comcast dot net (remove the spaces and do the & >>>obvious with the "dot" and the "at" >> >>Yes! >>G >>The Boot disk is limited to the RZ26 (1.07mb max size) but second and / >>third disks can be larger (but not bootable).  >>D >>My 3100s easily fit three drives if there is no Floppy/tk70/Crdom.B >>However with two or more drives make sure the fans work and the  >>dust is cleaned out. > E >I usually put them in external housings since I can move them around F >more easily. On top of the issue of finding the mounting trays if the >box didn't come with them... + >------------------------------------------ D >NO, use John Mee3 at comcast dot net (remove the spaces and do the $ >obvious with the "dot" and the "at"  D I have BA42 drive boxes to bump the number of drives to greater thanF box capactity.  One of the 3100/76s has around 15GB of drives.  Plenty of room there.   Allison    ------------------------------  + Date: Thu, 20 Apr 2006 12:25:52 -0500 (CDT) * From: sms@antinode.org (Steven M. Schweda)! Subject: Re: Free rz26s and rz28s 2 Message-ID: <06042012255197_202002C3@antinode.org>  ? > Does anybody remember what the DEC part number was for a SCSI G > controller that would support both tape and disk?  Trying to cover my ' > bases for when the ESDI drives die... , > ------------------------------------------E > NO, use John Mee3 at comcast dot net (remove the spaces and do the  % > obvious with the "dot" and the "at"   F    Does anyone remember there being one?  I have an Emulex UC07 in oneI old box (a UC08 would do disk and tape), and a CMD CQD223/TM in another.  H These sorts of cards appear on Ebay from time to time, but I've not seenG one go cheap.  For enough money to retire on, I'd consider parting with 1 a card of mine, but I don't expect such an offer.   H ------------------------------------------------------------------------  3    Steven M. Schweda               sms@antinode-org 4    382 South Warwick Street        (+1) 651-699-9818    Saint Paul  MN  55105-2547    ------------------------------    Date: 20 Apr 2006 08:00:00 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) 5 Subject: Re: How does a fix become a published patch? 3 Message-ID: <nRyCFx17spGJ@eisner.encompasserve.org>   y In article <44469E77.D3A82406@NeOaSrPtAhMlNiOnWk.net>, "David J. Dachtera" <djesys.nospam@NeOaSrPtAhMlNiOnWk.net> writes:  > G > Would violate any rules of propriety to give us a clue as to why that  > is?   C    Hm.  I wonder if there were enough bits in my MV II to track all -    those blocks on a large ODS-5 volume?  8-)   G    IIRC 64 bit addressing and/or 64 bit arithmetic have something to do A    with file system features not available on VAX.  I suppose one H    could slow down the file system with lots of EMUL and ADDC and littleF    subroutines to do all the rest of the 64 bit arithmetic but I can'tG    see slowing it down with some kind of mega cathedral window turns to      get around 64 bit addressing.   ------------------------------    Date: 20 Apr 2006 10:25:39 -0500. From: brooks@cuebid.zko.hp.nospam (Rob Brooks)5 Subject: Re: How does a fix become a published patch? , Message-ID: <lVoTJtn2TmMx@cuebid.zko.hp.com>  B "David J. Dachtera" <djesys.nospam@NeOaSrPtAhMlNiOnWk.net> writes: > JF Mezei wrote:  >> "David J. Dachtera" wrote:    >> > I wrote . . .  R >> > > This goes back to the minicopy project, where VAXes can contribute to writeP >> > > bitmaps that exist on Alpha systems, but VAXes themselves cannot have the= >> > > master copy, nor can they perform minicopy operations.   J >> > Would violate any rules of propriety to give us a clue as to why that >> > is?  K >> I think it was explained before. The VAX versions of VMS don't have full  >> support of minicopy.    > Why?  K I wasn't in VMS Engineering when the Write Bitmap project started (1999, I   think).   J A reasonable guess, however, is that given limited manpower, resources areO allocated according to what business management perceives as the greatest need.   K Doing the work on the VAX to fully support Write Bitmaps would have greatly M increased the resources needed for WBM within VMS Engineering -- it is not as K simple as recompile-and-go as part of the VAX build, and the time needed to L qualify and validate the VAX work would have been as large as that needed to qualify and validate on Alpha.    N HBMM makes use of Write Bitmaps, so HBMM on the VAX was never a consideration, given what I've written above.  L Like it or not, we constantly make tradeoffs.  I (personally) have about 10 O man-years worth of work in front of me, in terms of things that I'm working on, L expect to be working on, and hope to work on.  It's pretty much a given thatE I won't do all of the projects in the "expect" and "hope" categories.   O A classic example (for me) is "concurrent multipath" -- that is, sending I/O to J more than just one path at a time.  Now, only the single "current" path isI used. We think that we'd see an increase in I/O throughput if we used all D available paths on the active controller, although that has not been quantified.   L Note, however, I said "active controller"; that does not mean that we'd everJ use all available paths on (for example) an HSG80.  Only half of the totalN paths could be used concurrently.  A LUN can only be active to one half of theK redundant pair, and it's quite intrusive to yank the LUN from controller to  controller.   M In any even, concurrent multipath has been on my horizon for a few years, but $ other things have taken precedence.     --    H Rob Brooks    VMS Engineering -- Exec Group     brooks!cuebid.zko.hp.com   ------------------------------    Date: 20 Apr 2006 07:19:41 -07007 From: "refillkitarg@gmail.com" <refillkitarg@gmail.com> @ Subject: INK JET concentrate Manufacturer ariel@refillkit.com.arC Message-ID: <1145542781.342959.147440@t31g2000cwb.googlegroups.com>   7 INK JET concentrate Manufacturer ariel@refillkit.com.ar  RFKT CONCENTRATE BASE 4000 DESCRIPTION B We can offer you our product, RFKT CONCENTRATE BASE 4000, in BLACKG color, with only water  and UNDER OUR INSTRUCTIONS, you will be able to E manufacture your ink for any cartridges brand (HP - LEXMARK - CANON -  EPSON - XEROX). E The RFKT CONCENTRATE BASE 4000 is a chemical mixture of high quality, C manufactured by experts in the matter, with the aditionals of  only E water and one filter easy purchase, you can obtain production of ink, + only mixing and without any machine needed.  YIELD G WITH EACH 1 LITER of this product, you will be able to produce 4 LITERS 7 OF INK, reducing costs in shipping, taxes and volumens.  PRICE G RFKT CONCENTRATE BLACK BASE INK 4000 Price is U$S 24- FOR LITER and the  minimum purchase is 10 LITERS.C Color ink concentrate color base INK 4000 U$S 30- FOR LITER and the E minimum purchase is 10 (LITERS WITH EACH 1 LITER of this product, you . will be able to produce 5 LITERS OF COLOR INK)  
 ADVANTAGESC If you are outside Argentina and are interested in distributing our G products, this can be an excellent possibility to begin. With 10 Liters B of RFKT CONCENTRATE BASE 4000 , you will be able to manufacture 40B Liters of Ink for HP - EPSON - LEXMARK - CANON - XEROX. With  onlyF water and  one filter  of easy purchase in any city - you will be able: to manufacture. The necessary support will be offer by us.
 CONTACT US; If you are interested in this product, send us an e-mail to 1 ariel@refillkit.com.ar  www.refillkit.com.ar  MSN  ariel@refillkit.com.ar   ------------------------------  % Date: Thu, 20 Apr 2006 08:26:00 -0400 ' From: "Main, Kerry" <Kerry.Main@hp.com> N Subject: RE: Is OpenVMS certified yet for 64-way Hard partitions (GS1280 M64).T Message-ID: <FA60F2C4B72A584DBFC6091F6A2B8684012B66E4@tayexc19.americas.cpqcorp.net>   > -----Original Message-----J > From: David J. Dachtera [mailto:djesys.nospam@NeOaSrPtAhMlNiOnWk.net]=20 > Sent: April 19, 2006 4:51 PM > To: Info-VAX@Mvb.Saic.Com : > Subject: Re: Is OpenVMS certified yet for 64-way Hard=20 > partitions (GS1280 M64). >=20 > "Main, Kerry" wrote: > >=20  > > > -----Original Message-----  > > > From: David J. Dachtera=20/ > [mailto:djesys.nospam@NeOaSrPtAhMlNiOnWk.net] # > > > Sent: April 16, 2006 10:08 PM  > > > To: Info-VAX@Mvb.Saic.Com ; > > > Subject: Re: Is OpenVMS certified yet for 64-way Hard  > > > partitions (GS1280 M64). > > >  > > > "Main, Kerry" wrote: > > > > $ > > > > > -----Original Message-----B > > > > > From: Richard Maher [mailto:maher_rj@hotspamnotmail.com]' > > > > > Sent: April 15, 2006 11:46 PM # > > > > > To: Info-VAX@Mvb.Saic.Com ? > > > > > Subject: Re: Is OpenVMS certified yet for 64-way Hard " > > > > > partitions (GS1280 M64).	 > > > > >  > > > > > Hi David, 	 > > > > > F > > > > > > Not much to tell. Oracle has evolved from its VMS roots to > > > > > a UN*XlandA > > > > > > item. It runs within its own "realm", oblivious to=20  > O.S.-level > > > > > > clustering. 	 > > > > > C > > > > > Yes, yes, yes and I'm also led to believe that it prefers  > > > > > total control ofH > > > > > virgin disks that have no OS formatting or file system at all.	 > > > > >  > > > > @ > > > > That's UNIX stuff .. Oracle has always supported OpenVMS > > > file systems.  > > > A > > > Well, yes and no. It tolerates the environment, but gets=20  > hopelessly% > > > balled up on bound volume-sets.  > > >  > >=20G > > Well, in todays large SAN disk world, I suspect not many Cust's are H > > still using bound volumes, but in any event, this sounds like a bug. > > What was the error >=20/ > Irrecoverable database/filesystem corruption.  >=207 > > and what did Oracle Support state when you reported  > > it?  >=20& > Bound volume-sets are not supported. >=20  B Ok, so not supported means use bigger SAN volumes instead of boundC volumes. OpenVMS also allows SAN volumes to dynamically increase as F well, so is the issue of bound volumes really as big of an issue as it$ perhaps might have been in the past?  C > > > > In recent 9i and above, it doe require ODS5 file structures  > > > on OpenVMS.  > > > B > > > Not really a software requirement, but rather the product of > > > having been @ > > > developed in an environment where ODS-2's restrictions are > > > not the norm. @ > > > (Mixed case filenames, multiple dots, deep directories,=20
 > and so on.)  > > >  > >=20B > > ODS5 does provide the newer features that you stated, but I=20
 > am not sure  > > what the point was.  >=20< > That ODS-2 does not. Rather than re-work the product to=20 > observe ODS-2 asH > the "lowest common (filesystem) denominator", Oracle choose to exploitC > ODS-5's looser restrictions and live instead within those bounds.  >=20  8 Well, since Oracle's reference platform is based on UNIXB (Solaris/Tru64/Linux), if you were looking at ways to allow faster@ ports, would it not make sense to use the features of the target< platform that are closest to that of the reference platform?     [snip..]   Regards   
 Kerry Main Senior Consultant  HP Services Canada Voice: 613-592-4660  Fax: 613-591-4477  kerryDOTmainAThpDOTcom (remove the DOT's and AT)=20  4 OpenVMS - the secure, multi-site OS that just works.   ------------------------------    Date: 20 Apr 2006 06:26:07 -0700- From: "Andrew" <andrew_harrison@symantec.com> # Subject: Re: OT: Sparc not dead yet C Message-ID: <1145539567.437651.208720@t31g2000cwb.googlegroups.com>    JF Mezei wrote: e > http://news.com.com/Sun+at+work+on+Niagara%2C+Rock+successors/2100-1006_3-6062008.html?tag=nefd.top  > F > Sun announces successors to Niagara and Rock successors, at first at" > 90nm process and later at 65nm . >  > H > "Niagara has eight processing cores, each able to handle four threads,> > enabling a total of 32 independent software sequences to run9 > simultaneously. Niagara II increases this limit to 64;"  >  > 2 > IA64 is "soon" getting to 2 processors per core. >  > I > Looks to me like Sun carefully edged its bets, with both Sparc and 8086 I > and isn't going to prematurely staop Sparc with its huge installed base & > until the 8086 has really scaled up.  G The announcments from Sun thje Niagara II Tape out, Rock II and Niagara F III followons effectively indicate that Sun are committed to SPARC forF at least another 5-6 years. Anything could change in the mean time butF this is a pretty strong commitment and rather more than simply hedging	 its bets.   G The next couple of years are going to be very interesting. Sun have hit = the sweet spot with Niagara in terms of performance and power D consumption. The new EPA standards arround server environmentals mayB play straight into Sun's court making the T1 based systems a prime6 contender to replace x86 and Power/IA64 based systems.   > H > The big difference with IA64 is that IA64 has no installed base, it is- > PaRisc and Alpha that have installed bases.  > H > So those who were saying that Sparc was dead were wrong. It isn't deadF > yet, even if it might be true that Sun intends to eventually replace > Sparc with the 8086.  G There is no evidence that Sun ever seriously considered replacing SPARC G with x86 only speculation. What of course has changed is Sun's adoption D of x86 a strategy driven by the realisation that they can make money@ out of x86 platforms hosting Linux, Windows and Solaris plus theD adoption of a standard server packaging to support low end SPARC and x86 servers.   regards  Andrew Harrison    ------------------------------  + Date: Thu, 20 Apr 2006 08:54:07 +0000 (UTC) P From: helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply)E Subject: Re: Partitioned cluster question (reboot during lost quorum) $ Message-ID: <e27i7f$cp0$1@online.de>  C In article <1145449215.629422.111660@g10g2000cwb.googlegroups.com>,  etmsreec@yahoo.co.uk writes:    I > That said, I know of at least one system manager that specifically sets C > his clusters up with an EXPECTED_VOTES of 1 on every node so that G > should the network links fail and should they need to reboot one node G > all alone in the event of a failure that affects the other nodes they G > can do so.  I see his line of reasoning, but don't necessarily agree.   @ In that case, do a minimum boot.  This guy can get toasted by a  partitioned cluster.   ------------------------------  % Date: Thu, 20 Apr 2006 06:21:32 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> E Subject: Re: Partitioned cluster question (reboot during lost quorum) , Message-ID: <444760AB.9D0144FE@teksavvy.com>  / Phillip Helbig---remove CLOTHES to reply wrote: G > When all are there, it defines CLUSTER_UP in the cluster table.  When , > CLUSTER_UP is defined, mounting can begin.  F NOP !  In the event of a disaster where you need to turn a backup nodeG into production, you need to provide the service ASAP. In the case of a @ reboot where the backup node will now be alone because a nuclearF explosion occured a few blocks away at the production site, the backupH node would wait endlessly for the production node to appear, never mount, the disks and users never get service back !  H When I setup such a system, I had each node wait up to 5 minutes for allE shadowed disks to appear, but after that timeout, it would create the H shadowset with only the disks that were visible. But this is "dangerous"G and requires human thining on how/when to boot the nodes to ensure that F the right node containing data acceptable to the user is being booted.  E The use complained about the 5 minute wait, but I explained that if I @ mounted right away, it would then take 2-3 hours for a full copyD operation once the other drives would become online and get includedG into the shasowset (and during this time, system response time would be H sluggish ... User agreed that waiting 5 minutes was the better solution.   ------------------------------  % Date: Thu, 20 Apr 2006 06:15:29 -0400 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> E Subject: Re: Partitioned cluster question (reboot during lost quorum) , Message-ID: <44475F40.CD7BE9DC@teksavvy.com>  / Phillip Helbig---remove CLOTHES to reply wrote: K > > That said, I know of at least one system manager that specifically sets E > > his clusters up with an EXPECTED_VOTES of 1 on every node so that I > > should the network links fail and should they need to reboot one node     A > In that case, do a minimum boot.  This guy can get toasted by a  > partitioned cluster.  - Minimum boot won't give users any service....   H Not trying to defend this tactic, however in the event of a failure, theE node will still freeze due to loss of quorum. It would require manual B intervention to zap that node ( BREAK on OPA0 and then HALT at theD chevron)  and reboot it. One would *ASSUME* that the human, prior toE taking the decision to reboot that node, would have carefully studied 4 the situation to ensure no partitioning is possible.   ------------------------------  + Date: Thu, 20 Apr 2006 11:37:48 +0000 (UTC) P From: helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply)E Subject: Re: Partitioned cluster question (reboot during lost quorum) $ Message-ID: <e27rqc$q72$1@online.de>  5 In article <44475F40.CD7BE9DC@teksavvy.com>, JF Mezei ' <jfmezei.spamnot@teksavvy.com> writes:    1 > Phillip Helbig---remove CLOTHES to reply wrote: M > > > That said, I know of at least one system manager that specifically sets G > > > his clusters up with an EXPECTED_VOTES of 1 on every node so that K > > > should the network links fail and should they need to reboot one node  >  > C > > In that case, do a minimum boot.  This guy can get toasted by a  > > partitioned cluster. > / > Minimum boot won't give users any service....   A Minimum boot, change VAXCLUSTER to 0 or reduce EXPECTED_VOTES or   whatever then reboot.    ------------------------------  # Date: Thu, 20 Apr 2006 14:27:12 GMT , From: Hoff Hoffman <hoff-remove-this@hp.com>E Subject: Re: Partitioned cluster question (reboot during lost quorum) 2 Message-ID: <4TM1g.6582$Bt3.4752@news.cpqcorp.net>   JF Mezei wrote:  > Hoff Hoffman wrote: G >>    If you want to boot outside the cluster, then I tend to prefer to > >> avoid enabling NISCS_LOAD_PEA0 and I don't load VAXCLUSTER. > G > No, the idea is to be able to boot one node first before you bring in J > the other ones, so that first one has to form the cluster and the others > join in later.  F    Welll, my own idea is to be able to boot distinctly and completely A and fully outside the cluster, and to thus reduce the chances of  " creatively configured corruptions.   > K >>    Consider the last time this has happened, too.  I've been at this for D >> over a decade, and I can't recall ever having run into this case. > D > What I've come to realise is that it is very hard to predict every > possible problem/situation.   D    That's a central treatise of Disaster Tolerance (DT).  It's also E understood that small failures or critical procedures can derail the   entire effort.  J > The ability to find solutions out of some unpredicted situation requiresJ > good understanding of the clustering process and full understanding yourG > configuration/applications so that you can "cheat" the config without  > jeoperdizing data integrity.  I    My goal is to win with the least effort and the least risk, and -- as  H a rule, I prefer to configure and operate the configuration such that I I don't have to get "creative".  I'd prefer to win by avoiding the battle,  H because battles are expensive and wasteful.  (Yes, I've been re-reading I a translation of a very old book, but I digress.)  If I can configure to  H avoid the problem (through hardware or software upgrades, or documented 2 management procedures and sequences), then I will.    G > VMS cannot make any assumptions when it loses connection with another G > node. It could be ethernet down, it could be the other node down. VMS J > (rightly)  protects against worse case scenario.  But the system managerI > can obtain additional information (such as confirmation that the others D > nodes are in fact powered down) which then allows him to gauge the? > situation and consider  that the first node can be up without I > jeoperdizing data integrity, at which point knowledge on how to "cheat" H > the predefined rules comes in handy to get out of that situation. (and# > later restore the safe settings).   H    You're definitely headed toward DT here, and DT is far more involved H than it looks and that most folks initially assume.  I've been doing DT G for a number of years both in the technology area and in the emergency  F services area, and no plan I've seen has fully survived a disaster -- C some have done better than others (and the core infrastructure has  > stayed available), but some plans have been colossal failures.    Q >>  As a  rule, the node with the valid copy will have the highest instantiation, 8 >> and will be the source for the shadow copy operation, > H > Your "as a rule" applies well when all disks are local and if you haveJ > access to one disk, you have access to all disks. (aka, you always mountG > all members of a shadowset together in the same mount command, so the H > shadowing software can make the proper determination of which physical& > disk has the more recent valid copy. > E > But when shadowset members appear gradually as nodes boot, then the J > order in which you boot the nodes becomes critical to ensuring the rightI > physical drive is first mounted into the shadowset.  Knowledge of which J > physical drive has the valid contents comes from knowing the config, theJ > exact state of the cluster before the mishap and its current state. AKA:J > full situation awareness. And it isn't something which VMS can ascertain8 > by itself. It takes a human to diagnose the situation.  G    You haven't hit a case here where the automatic recovery will fail,  I BTW.  If the cluster configuration is correct, then the recovery will be   correct.  H    Again, how often is the case you are looking at likely to arise, and $ are there ways to avoid it entirely?  F > So knowing how expected_votes works, means you can cheat the defaultJ > config to allow one node to boot first when you know this is what has to
 > be done.  F    Cheating is not without its costs and its risks, and the penalties F can be severe.  I prefer to configure and to operate in a manner that G avoids the need for configurational creativity, particularly as forced  F creativity during the usual chaos can cause its own class of problems.  <    DT configurations do inherently depend on voice and data H communications links, and the critical "fail-over" decision -- to split H the lobes and to punt the processing in the affected lobe(s) -- is left K up to the operator(s) in most every full DT configuration I've worked with.   G    If you want or believe you need to perform a sequence as you appear  F to want to and if you believe you must bypass the blade guards and if H there is no other more supportable means available to either prevent or B to provide for or to avoid the requirement, then you will want to I practice and document and test and carefully follow the sequence, as you  H will want to make every effort to avoid the corruptions that can ensue. G   If you can avoid it, you don't want to require yourself or others --  I and do remember that your creativity may not be fully appreciated or may  C not be correctly executed by others -- to become reactive within a   difficult situation.   ------------------------------  # Date: Thu, 20 Apr 2006 14:29:45 GMT , From: Hoff Hoffman <hoff-remove-this@hp.com>E Subject: Re: Partitioned cluster question (reboot during lost quorum) 1 Message-ID: <tVM1g.6583$Bt3.816@news.cpqcorp.net>   / Phillip Helbig---remove CLOTHES to reply wrote: E > In article <1145449215.629422.111660@g10g2000cwb.googlegroups.com>,  > etmsreec@yahoo.co.uk writes:   > J >> That said, I know of at least one system manager that specifically setsD >> his clusters up with an EXPECTED_VOTES of 1 on every node so thatH >> should the network links fail and should they need to reboot one nodeH >> all alone in the event of a failure that affects the other nodes theyH >> can do so.  I see his line of reasoning, but don't necessarily agree. > B > In that case, do a minimum boot.  This guy can get toasted by a  > partitioned cluster.  C    Ayup.  One of the local engineers accidentally booted into that  H configuration, with a simple console command mistake and mis-set voting H parameters and a shared SCSI.  The corruptions that resulted were quite  impressive.    ------------------------------    Date: 20 Apr 2006 09:54:09 -0500 From: briggs@encompasserve.orgE Subject: Re: Partitioned cluster question (reboot during lost quorum) 3 Message-ID: <8y3gKkBNFrd+@eisner.encompasserve.org>   w In article <e27rqc$q72$1@online.de>, helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply) writes: 7 > In article <44475F40.CD7BE9DC@teksavvy.com>, JF Mezei ) > <jfmezei.spamnot@teksavvy.com> writes:   > 2 >> Phillip Helbig---remove CLOTHES to reply wrote:N >> > > That said, I know of at least one system manager that specifically setsH >> > > his clusters up with an EXPECTED_VOTES of 1 on every node so thatL >> > > should the network links fail and should they need to reboot one node >>   >>  D >> > In that case, do a minimum boot.  This guy can get toasted by a >> > partitioned cluster.  >>  0 >> Minimum boot won't give users any service.... > C > Minimum boot, change VAXCLUSTER to 0 or reduce EXPECTED_VOTES or   > whatever then reboot.   L Then you're (probably) not talking about minimum boot.  You're talking about a conversational boot.  D The usual way to request a minimum boot is by doing a conversational% boot and setting STARTUP_P1 to "MIN".    ------------------------------  # Date: Thu, 20 Apr 2006 15:17:53 GMT 1 From: Keith Parris <keithparris_NOSPAM@yahoo.com> 3 Subject: Re: Quorum, locks and application question 2 Message-ID: <BCN1g.6585$fy3.4056@news.cpqcorp.net>   Richard Maher wrote: > Consider this conundrum: - > I > Q. How does a Rdb Monitor process faithfully implement it's FREEZE lock K > mechanism so as to protect the ACID properties of a database transaction?  > J > A. Stuffed if I know. (There was a discussion in ITRC about this a while > back)  > L > 1) Process A on Node X has entered a Cayman Island transaction for $1B andH > will have to rollback 'cos a commit-time constraint will stop it going
 > through.L > 2) Process B on Node Y has a lock pending on this Account record and wants- > to transfer the balance to the Isle of Man. K > 3) Node Z is thrashing away sending tree re-mastering requests all around * > the cluster and being a general nuisance > 4) Node X dies > : N > : From this point on I imagine the possibility of a race-condition. That is,D > what *right-now* is stopping Process B from seeing the dodgy data?  F If I understand your question, the basic issue is that when a lock is F being used to protect updates to some shared resource, you may really G need to know if the reason you are now being granted your lock request   is because: G 1) the prior user is now done with it, having left the shared resource  1 (like a database block) in a consistent state, or G 2) the prior user got interrupted in mid-update by a node failure, and  H the resource has been left in an indeterminate state and probably needs  to be fixed up  H In the early '90s Rdb asked for and received new capabilities in VMS to I handle this possibility. The $SETCLUEVT system service allows one to ask  I for notification of events like loss of a node from the cluster, and VMS  F guarantees that the AST notifying you of the node loss will always be G delivered before any AST (or mainline process execution) notifying you  G that a has lock granted because the old lock on a failed node has been  * freed as a result of the state transition.   ------------------------------  # Date: Thu, 20 Apr 2006 15:26:56 GMT 1 From: Keith Parris <keithparris_NOSPAM@yahoo.com> 3 Subject: Re: Quorum, locks and application question 2 Message-ID: <4LN1g.6586$iw3.6469@news.cpqcorp.net>   JF Mezei wrote: J > If a node-A is disconnected from ethernet and , from its point of view ,J > loses quorum and freezes, from the point of view of the other nodes, the; > locks held by applications on node-A will vanish. Right ?   H The other nodes have to wait for RECNXINTERVAL seconds before they give E up on A, initiate a state transition, and free the locks it formerly  A held. During this time, quorum is effectively lost on node A, so  H processes on A can't be scheduled for execution, because by default all I processes require the QUORUM capability bit, and when quorum is lost the  C QUORUM capability bit is removed from all CPUs, so there is no CPU  H eligible to run the processes. (In ancient days, VMS looped at IPL 4 on  a loss of quorum.)  J > But when Node-A is reconnected, the expected behaviour is that the otherI > nodes will tell Node-A to commit suicide.Does this happen before normal 1 > applicatiosn on Node-A are given any CPU time ?   F Yes. Node A never regains quorum. It just gets the bad news that it's H been kicked out of the cluster (because the locks it held are no longer 0 valid) and reboots so it can rejoin the cluster.  G > In other words, if I have an application on Node-A, will the last CPU G > cycle its gets happen just before quorum is lost, and the application G > will get NOTHING  between the time the connection is restored and the D > time Node-A accepts the other nodes' request to commit hara-kiri ?   Right.  J > Or would the application have some opportunity to run some exit handlers/ > in the process of Node-A committing suicide ?   H No. No processes can be scheduled for execution while quorum is lost on  a node.   G > (Looking at the lock handling, I can really understand now why Node-A F > has to commit suicide. Once othert nodes discard of locks held by A,@ > they can then take those locks. And if A is allowed to resume,J > applications on A will continue to think that they are the holder of the3 > lock when in fact it is the other nodes that do).    Exactly.   ------------------------------  % Date: Thu, 20 Apr 2006 13:21:25 -0400 ' From: Dave Froble <davef@tsoft-inc.com> 3 Subject: Re: Quorum, locks and application question / Message-ID: <XpSdnZIT7p0NX9rZRVn-rw@libcom.com>    Keith Parris wrote:  > Richard Maher wrote: >  >> Consider this conundrum: -  >>J >> Q. How does a Rdb Monitor process faithfully implement it's FREEZE lockL >> mechanism so as to protect the ACID properties of a database transaction? >>K >> A. Stuffed if I know. (There was a discussion in ITRC about this a while  >> back) >>J >> 1) Process A on Node X has entered a Cayman Island transaction for $1B  >> andI >> will have to rollback 'cos a commit-time constraint will stop it going  >> through. H >> 2) Process B on Node Y has a lock pending on this Account record and  >> wants. >> to transfer the balance to the Isle of Man.L >> 3) Node Z is thrashing away sending tree re-mastering requests all around+ >> the cluster and being a general nuisance  >> 4) Node X dies  >> :G >> : From this point on I imagine the possibility of a race-condition.   >> That is, E >> what *right-now* is stopping Process B from seeing the dodgy data?  >  > H > If I understand your question, the basic issue is that when a lock is H > being used to protect updates to some shared resource, you may really I > need to know if the reason you are now being granted your lock request  
 > is because: I > 1) the prior user is now done with it, having left the shared resource  3 > (like a database block) in a consistent state, or I > 2) the prior user got interrupted in mid-update by a node failure, and  J > the resource has been left in an indeterminate state and probably needs  > to be fixed up > J > In the early '90s Rdb asked for and received new capabilities in VMS to K > handle this possibility. The $SETCLUEVT system service allows one to ask  K > for notification of events like loss of a node from the cluster, and VMS  H > guarantees that the AST notifying you of the node loss will always be I > delivered before any AST (or mainline process execution) notifying you  I > that a has lock granted because the old lock on a failed node has been  , > freed as a result of the state transition.  E So if a lock is granted because the node holding the lock exited the  A cluster, the process acquiring the lock must be testing for this  F condition by having such an AST queued every time a lock is requested?  I Does RDB queue an AST and test for the condition.  (I'm guessing that it  G does since your post is in reply to such an event.)  In the case where  G the condition exists, what does RDB do?  It's like "hey buddy, you now  < have the resource, but the last guy never finished with it".   --  4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com 170 Grimplin Road  Vanderbilt, PA  15486    ------------------------------    Date: 20 Apr 2006 07:42:29 -0700 From: glen.thompson@gmail.com ' Subject: Recommended sources for cables B Message-ID: <1145544149.637062.15040@i40g2000cwc.googlegroups.com>  B I'm in need of a few BN24H cables for an upcoming equipment/officeB move.  Does anyone have a source that they recommend?  Looking for2 decent price, reliable service, prompt delivery.     TIA, glen   ------------------------------    Date: 20 Apr 2006 07:56:14 -0700; From: "johnhreinhardt@yahoo.com" <johnhreinhardt@yahoo.com> + Subject: Re: Recommended sources for cables C Message-ID: <1145544974.643464.327600@t31g2000cwb.googlegroups.com>    glen.thompson@gmail.com wrote:D > I'm in need of a few BN24H cables for an upcoming equipment/officeD > move.  Does anyone have a source that they recommend?  Looking for2 > decent price, reliable service, prompt delivery. >  > TIA, > glen  ( I've had good luck with Stonewall Cable.D http://www.stonewallcable.com/  Prices seem reasonable.  You can getF the cables in standard lengths or custom. Service is pretty fast. I'veB placed a couple orders.  Longest took about a week with UPS ground delivery to Ohio.      John H. Reinhardt    ------------------------------  % Date: Thu, 20 Apr 2006 11:11:49 -0400 C From: "David Turner, Island Computers US Corp" <dbturner@icusc.com> + Subject: Re: Recommended sources for cables 8 Message-ID: <cvN1g.6257$Kn4.1176@bignews2.bellsouth.net>   We have them in stock NEW    David        --     David B Turner Island Computers US Corp 2700 Gregory St, Suite 180 Savannah GA 31404  Tel: 912 447 6622 X201 Cell: 912 447 6622 X251  Fax: 912 201 0402  Email: dbturner@icusc.com  Web: http://www.islandco.com% ===================================== < All orders are subject to the following terms and conditions. of sale. These should be read before ordering.% http://www.islandco.com/warranty.html   * <glen.thompson@gmail.com> wrote in message< news:1145544149.637062.15040@i40g2000cwc.googlegroups.com...D > I'm in need of a few BN24H cables for an upcoming equipment/officeD > move.  Does anyone have a source that they recommend?  Looking for2 > decent price, reliable service, prompt delivery. >  > TIA, > glen >    ------------------------------  # Date: Thu, 20 Apr 2006 14:31:34 GMT ; From: "Jeffrey H. Coffield" <jeffrey@digitalsynergyinc.com>  Subject: Sending mail via TLS > Message-ID: <aXM1g.50534$_S7.25566@newssvr14.news.prodigy.com>  H One of my customers needs to send e-mail from Alpha VMS 7.3-2 using TLS.C I see that PMDF from Process Software does this but are there other 
 solutions?   TIA 
 Jeff Coffield    ------------------------------    Date: 20 Apr 2006 05:59:30 -0700 From: Bart.Zorn@gmail.com 7 Subject: Re: Strange behavior of DEFINE with equal-sign C Message-ID: <1145537970.256146.161290@i40g2000cwc.googlegroups.com>    Bob Koehler wrote:r > In article <OF085482AF.556870CE-ON85257155.006DD6F1-85257155.0071371B@metso.com>, norm.raphael@metso.com writes: > >  > > ..then how do you reconcile  > > $ name="Doug2 > > $ write sys$output "Hello, my name is ''name'! > > H > > which as two instances of single double-quotes (bug which works) andJ > > one instance of double single-quote/single single-quote (an odd numberH > > which you would put the author into the group above with bad hygene)& > > which looks wrong but is required! > ; >    'name and &name are different, but similar operations.  > 0 >    'name' is not the same operation as 'name .  ' Care to elaborate about the difference?    > 2 >    ''name' does both the ' and the '' operation.  > Hmm. Assuming that we are talking about the occurrence of thisE construct outside a quoted string, this casts a heavy shadow over the  previous statement!    > I >    "string<EOL> is the only undocumented feature I see in your example.  > 9 >    Bad hygene is probably a sign of a eunichs' problem.    Regards,  	 Bart Zorn    ------------------------------    Date: 20 Apr 2006 07:51:11 -0500; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) 7 Subject: Re: Strange behavior of DEFINE with equal-sign 3 Message-ID: <cIDXj8wbQUvc@eisner.encompasserve.org>   p In article <OF085482AF.556870CE-ON85257155.006DD6F1-85257155.0071371B@metso.com>, norm.raphael@metso.com writes: >  > ..then how do you reconcile  > $ name="Doug0 > $ write sys$output "Hello, my name is ''name'! > F > which as two instances of single double-quotes (bug which works) andH > one instance of double single-quote/single single-quote (an odd numberF > which you would put the author into the group above with bad hygene)$ > which looks wrong but is required!  9    'name and &name are different, but similar operations.   .    'name' is not the same operation as 'name .  0    ''name' does both the ' and the '' operation.  G    "string<EOL> is the only undocumented feature I see in your example.   7    Bad hygene is probably a sign of a eunichs' problem.    ------------------------------    Date: 20 Apr 2006 08:29:14 -0700% From: "Laura" <lmcgaughey@parsec.com> ! Subject: The Undocumented OpenVMS C Message-ID: <1145546954.092806.210310@g10g2000cwb.googlegroups.com>    Hello,  < I'd like to invite you to sign up for our FREE Webinar, "TheD Undocumented OpenVMS," to be held on Wednesday, May 17 at 12:00 noon MDT.  = This one-hour Webinar will cover the undocumented features of E OpenVMS--some humorous, some useful, and some you might already know. A Who knows where this hour may take us? Come join us in the fun in 7 discovering and discussing these features and find out!   < You can register for "The Undocumented OpenVMS" by going to:  4 http://www.parsec.com/general/promotion.php?p=86O14L  = Thank you! Please feel free to contact me should you have any 
 questions.   Laura McGaughey  PARSEC Group Direct: 720-962-9583 Toll Free: 888-472-7732  Fax: 303-763-9909  lmcgaughey@parsec.com  www.parsec.com  ; Check out our partner site for news, jobs, and resources at  www.openvmsplanet.org!   ------------------------------   Date: 20 Apr 2006 10:16:43 GMT7 From: yehavi@vms.huji.ac.il (Yehavi Bourvine (58-4279)) $ Subject: VAX hardware error code 3 ?% Message-ID: <2006Apr20.101643@hujicc>    Hello,  N   I have a VAX-4,300 which failed; the LED display shows 3. Any idea what thisL error means, or where I can find the right documentation? I couldn't find it# neither on HP's site nor on Google.   9                                         Thanks! __Yehavi:    ------------------------------    Date: 20 Apr 2006 01:25:12 -07003 From: "Ralf Gaertner" <ralf.gaertner@t-systems.com> ( Subject: Re: VAX hardware error code 3 ?C Message-ID: <1145521512.144649.247890@g10g2000cwb.googlegroups.com>   ' The following manuals might be helpful:   7 EK-335AC-IN-003 VAX 4000 (Model 300) Installation Guide . (http://decdoc.itsx.net/dec94mds/335adin3.pdf)  5 EK-336AC-OP-003 VAX 4000 (Model 300) Operators Manual . (http://decdoc.itsx.net/dec94mds/336acop3.pdf)  @ EK-337AB-TI-002 VAX 4000 (Model 300) Technical Information Guide. (http://decdoc.itsx.net/dec94mds/337abti2.pdf)  9 EK-386AB-TS-002 VAX 4000 Model 300 Troubleshooting Manual . (http://decdoc.itsx.net/dec94mds/386abts2.pdf)  5 I didn't check the manuals for your specific problem.    Ralf   ------------------------------  # Date: Thu, 20 Apr 2006 16:56:12 GMT , From: Hoff Hoffman <hoff-remove-this@hp.com>G Subject: Re: VAX Q-bus and SCSI Disks?  (was: Re: Free rz26s and rz28s) 2 Message-ID: <M2P1g.6591$pz3.4208@news.cpqcorp.net>   foo@bar.com wrote:  ? > Does anybody remember what the DEC part number was for a SCSI G > controller that would support both tape and disk?  Trying to cover my ' > bases for when the ESDI drives die...   F    The KZQSA was not supported for magnetic disks, though it's likely A the option you remember -- the option reportedly worked for some  F magnetic disks, and failed for others.  It was supported for specific ! tapes and specific optical disks.   G    The path from Q-bus to SCSI that was supported involved a KFQSA and  G an HSD-series adapter, and you'll see this (or possibly the integrated  G DSSI and the HSD-series widget) on many of the later Q-bus VAX systems.   G    Old DIGITAL Systems and Options Catalog (SOC) entries are posted in  I the left navigation over at <http://www.hp.com/go/productbulletin/>, and  D you should be able to find both the KZQSA and the KFQSA specs there.   ------------------------------  % Date: Thu, 20 Apr 2006 05:25:11 -0700 ( From: Jeff Cameron <roktsci@comcast.net>. Subject: Re: What (WAS) wrong with my network?0 Message-ID: <C06CCBB7.1E62B%roktsci@comcast.net>  I On 4/19/06 6:46 AM, in article 000b01c663b7$9f004d80$994614ac@domina.fom, ) "Rudolf Wingert" <win@fom.fgan.de> wrote:    > Hello,H > Jeff, I am right, you don't have a manageable switch? But it should beI > an additional problem. If I did red your email right, you see different J > throughput between the Alphas with 100Base-TX in condition of direction.H > This looks strange. May be, there will be another problem too. Did you= > check the parameters (PIPELINE QUOTA, RECIEVE BUFFERS,...)?  > Best regards R. Wingert  > H Yes, My Pipeline quota is fine and My receive buffers are set to 32 (theL max). Plus the counters have a zero value in buffer unavailable. When I haveJ full duplex on, the Pele to AXPAXP traffic sends only one packet every 1.5 seconds.   ------------------------------   End of INFO-VAX 2006.219 ************************