INFO-VAX Thu, 15 Nov 2007 Volume 2007 : Issue 625 Contents: Re: "Latch" style Batch Queue Re: Hardcopy documentation Re: Hardcopy documentation Re: Need help with TCPIP V5.1 configuration Re: Need help with TCPIP V5.1 configuration Re: SYS$GRANT_LICENSE output codes Re: SYS$GRANT_LICENSE output codes system constants in COBOL Re: system constants in COBOL Re: system constants in COBOL Re: system constants in COBOL Re: TCPIP SMTP and/or BIND Resolver bug Re: TCPIP SMTP and/or BIND Resolver bug Re: TCPIP SMTP and/or BIND Resolver bug VIOC: sizing and performance RE: VIOC: sizing and performance ---------------------------------------------------------------------- Date: Wed, 14 Nov 2007 17:46:09 -0700 From: "Michael D. Ober" Subject: Re: "Latch" style Batch Queue Message-ID: <13jn5mmlt9h9r4f@corp.supernews.com> David, Thanks. I'll see if I can work from your example. If not, I'll be back with what I have tried. Mike. "David J Dachtera" wrote in message news:47367FAC.9F9670D4@spam.comcast.net... > David J Dachtera wrote: >> >> "Michael D. Ober" wrote: >> > >> > Is there a way in VMS 8.3 to create a queue that when it is started, >> > will >> > only process the jobs in the queue at the time it's started, but then >> > hold >> > any additional jobs that are submitted until it starts again. >> > Basically, I >> > need to create a queue that will accumulate jobs during the day and >> > then >> > start after the system completes database cleanups at night and then >> > immediately stop. >> >> I'd tend to agree with those who suggest having a Generic Batch queue >> with no >> /GENERIC_TARGETS and no execution queues which accept jobs from generic >> queues >> (no /ENABLE_GENERIC on any execution queues). >> >> SUBMIT your jobs to that queue during the processing day. >> >> During your nightly procedures, ASSIGN/QUEUE/MERGE that queue to another >> Generic >> queue which has at least one execution queue as a /GENERIC_TARGET. You >> can use a >> WAIT loop and the PENDING_JOB_COUNT item of F$GETQUI() to wait until the >> first >> queue has "drained" into the second, then DEASSIGN/QUEUE the "holding" >> queue so >> it will accept new jobs and hold them until you again run the job to >> drain it >> into the other queue. >> >> I know - clear as mud, eh? > > Actually, having tried it here on my little Alpha, all you would really > need > might look like this: > > $ INIT/QUE/BATCH/START OVERNIGHT_EXECUTOR1 > $ INIT/QUE/BATCH/START OVERNIGHT_EXECUTOR2 > $ INIT/QUE/BATCH/START OVERNIGHT_EXECUTOR- > /GENERIC=(OVERNIGHT_EXECUTOR1,OVERNIGHT_EXECUTOR2) > $ INIT/QUE/BATCH OVERNIGHT_TASKS/GENERIC > > That sets it up. > > Set up the jobs that run during the course of the day (on other queues) to > SUBMIT overnight processing to the OVERNIGHT_TASKS queue. > > Then, at the time of your choosing, have a scheduled job run a command > like > this: > > $ ASSIGN/MERGE OVERNIGHT_EXECUTOR OVERNIGHT_TASKS > > That will move all of the jobs queued up on the OVERNIGHT_TASKS queue to > the > OVERNIGHT_EXECUTOR queue, which will then distribute jobs to the > OVERNIGHT_EXECUTOR1 and OVERNIGHT_EXECUTOR2 queues. On a two-node cluster, > one > queue on each node would help distribute the load. > > The OVERNIGHT_TASKS queue should always remain STOPped if you have any > batch > execution queues which are INITIALIZEd using /ENABLE_GENERIC (which is the > default). > > -- > David J Dachtera > dba DJE Systems > http://www.djesys.com/ > > Unofficial OpenVMS Marketing Home Page > http://www.djesys.com/vms/market/ > > Unofficial Affordable OpenVMS Home Page: > http://www.djesys.com/vms/soho/ > > Unofficial OpenVMS-IA32 Home Page: > http://www.djesys.com/vms/ia32/ > > Unofficial OpenVMS Hobbyist Support Page: > http://www.djesys.com/vms/support/ > ------------------------------ Date: Wed, 14 Nov 2007 19:44:11 -0500 From: "John Smith" Subject: Re: Hardcopy documentation Message-ID: <92f2$473b963e$cef8877a$28301@TEKSAVVY.COM-Free> VAXman- @SendSpamHere.ORG wrote: > In article , "Robert > Jarratt" writes: {...snip...} >> Must be my poor memory about the 8 volumes then, or was it that the >> system calls was volume 8? It is hard to remember back 20+ years. >> Please do let me > > With age comes something but I can't remember what! :) > > > >> know what you have and how much it would cost, but I have to say >> that funds are limited and if it is too much I may have to pass, or >> perhaps just have a subset if you are prepared to do this. > > If I could find it in my heart to part with my set, I'd let you have > it. > I've held onto the old manuals as it's a sort of personal > d|i|g|i|t|a|l > museum. I did part with some stuff simply because I needed some > breath- > ing room some time ago. (uVAX-II, 9 track, Pro-350, etc., etc.) > IIRC, > Bob Koehler and a friend of his came to inherit some of this > memorabilia. > > I gave a couple of VAXen (3100 series) and Alphas (AS200) away too to > a > few "hobbyists". They really had no value for sale. I asked only to > be > reimbursed the shipping costs. Glad to hear you are out of the hospital. FYI - as to your manuals and wanting to keep a copy of them, I came across a very good personal scanner last week - Fujitsu Scansnap S510. It comes in two models, S510 for PC's, and S510M for Mac. It's rated at 18 page per minute BOTH sides simultaneously, for a total of 36 images per minute, and will OCR as it creates the .pdf files In the demo I saw the real speed was more like 13 ppm double-sided, which is damn good for a $400 scanner. But that's just the paper speed - the OCR takes several minutes longer. I'm in the midst of clearing my office of lots of old paper files and non-DEC manuals and think so highly of this scanner that I'll be getting one. My Brother multifunction scanner is ok but sucks when it comes to double-sided because I have to manually flip the pages to scan the 2nd side, and the software selection for duplexing the document is has to be manually triggered, and the OCR is not automatic. Amazon has these Fujitsu scanners, and in Canada www.ats-systems.com is where I'd recommend. Fujitsu also has a rebate this months for $50 in both Canada and the US. For higher volume scans, Fujitsu and Canon have good duplex models that will scan 30-50+ pages/minute and OCR at almost the same speed, in the mid-$2-3000 range. -- OpenVMS - The never-advertised operating system with the dwindling ISV base. ------------------------------ Date: Thu, 15 Nov 2007 01:34:47 GMT From: VAXman- @SendSpamHere.ORG Subject: Re: Hardcopy documentation Message-ID: In article <92f2$473b963e$cef8877a$28301@TEKSAVVY.COM-Free>, "John Smith" writes: {...snip...} > >Glad to hear you are out of the hospital. :( ... but another big doctor day tomorrow. I really only have these manual for historic purposes. I seldom look at them since I'm well beyond V4, V5 and V6 these days. Anytime I do need to reference docs today, I do so with the on-line docs. -- VAXman- A Bored Certified VMS Kernel Mode Hacker VAXman(at)TMESIS(dot)COM "Well my son, life is like a beanstalk, isn't it?" http://tmesis.com/drat.html ------------------------------ Date: Thu, 15 Nov 2007 00:56:46 +0100 From: ss Subject: Re: Need help with TCPIP V5.1 configuration Message-ID: Hi... An update: Thanks for the idea with the batch-job, well i had to initialize the batch-queue first but this was on my agenda anyway. No output, sorry, iīve forgot to c&p that part. Nevermind, the point is, that it proved that the system as a whole was affected, no input possible anymore, had to turn off the emulation as usual. Well, i wanted to do the rest of the investigations tomorrow, but couldnīt avoid taking a look at some of the TCPIP$.*COM files, where TCPIP$INET_STARTUP.COM sounded very insteresting to me. So $TYPE /PAGE TCPIP$INET_STARTUP.COM and i think iīve found the relevant part: [TCPIP$INET_STARTUP.COM] [...] $ ! $ ! load proxy execlet $ ! $ load_proxy: $ define /system/executive_mode/nolog - tcpip$proxy_services - sys$loadable_images:tcpip$proxy_services.exe $ $ sysconfig := $sys$system:tcpip$sysconfig.exe $ sysconfig -c proxy $ ! [...] Letīs try it out: $ RUN SYS$LOADABLE_IMAGES:TCPIP$PROXY_SERVICES.EXE %SYSTEM-F-ACCVIO, access violation, reason mask=04, virtual address=7FFE5E78, PC=00009479, PSL=03C00000 Improperly handled condition, image exit forced. Signal arguments Stack contents Number = 00000005 00000000 Name = 0000000C 20FC0000 00000004 7FEC6BFC 7FFE5E78 7FEC6BE8 00009479 7FF59657 03C00000 00000004 7FF592FC 7FFE5E0C 7FFE5EE4 7FFED188 Register dump R0 = 7FFE5E40 R1 = 00009800 R2 = 00000004 R3 = 7FF592FC R4 = 7FFE5E3C R5 = 7FFE5EE4 R6 = 00009400 R7 = 7FFED188 R8 = 7FFECA48 R9 = 7FFECC50 R10= 7FFED7D4 R11= 7FFE2BDC AP = 7FEC6B64 FP = 7FEC6B24 SP = 7FEC6BA0 PC = 00009479 PSL= 03C00000 Could this have hung the whole system? Iīll google and try some logging tomorrow, maybe i find out more... -Sascha In article , JF Mezei wrote: > ss wrote: > > > %TCPIP-I-STARTNAME, starting name service > > %%%%%%%%%%% OPCOM 12-NOV-2007 16:46:55.48 %%%%%%%%%%% > > Message from user Proxy Server on SIMVAX > > Loading proxy server image TCPIP$PROXY_SERVICES > > > Ok, looking at the startup procedure, it generates temporary files that > contain the services that need to be activated. One of then contains an > ordered list. It seems to start with bind, proxy and portmapper. (they > are hardcoded). The rest is added from configuration data. > > TCPIP> GENERATE SERVICE/CONFIGURATION/FILE=temp.txt > and then DUMP TEMP.TXT and you will see the services it will attempt to > start. > > What you could do is to not start TCPIP in your system startup > procedure. Once booted, you login interactively, and submit a batch job > to do the TCPIP$SYSTARTUP. (or spawn/nowait/output=log.file @tcpip$startup). > > While it is running, you can do a DIR disk:[SYS0.SYSTARTUP] for any > TCPIP*.TMP or TCPIP*.DAT , one of them will contain the ordered service > startup list. You know it has done the proxy startup, you might wish to > find out which is the next service on the list it was trying to start > when it hung. > > Also, if your interactve session doesn't hang, you will know it is only > the TCPIP$STARTUP that hangs (as opposed to the whole system hanging). > > you can also SHOW LOG /TABLE=TCPIP$STARTUP_TABLE and it shows which > services have been started, by whom and at what time. > > I would have to delve further in the TCPIP$STARTUP.COM file. The later > also does an @TCPIP$CONFIG GENERATE_SERVICES which is an undocumented > function. ------------------------------ Date: Wed, 14 Nov 2007 20:04:51 -0500 From: JF Mezei Subject: Re: Need help with TCPIP V5.1 configuration Message-ID: <64402$473b9b36$cef8887a$1680@TEKSAVVY.COM> I looked at my tcpip$STARTUP.LOG and if you looked at yours, it would probably show exactly where it stopped. After starting the proxy service, it starts routing. You may wish to inestigate your routing config. In particular, TCPIP SHOW ROUTE as well as the type of routing you chose in the TCPIP$CONFIG (no need for routed or gated for simple You may need to create your routing database in particular, TCPIP CREATE ROUTE and TCPIP CREATE NETWORK I recall problems with routing not happy that the network database was empty or something to that order. ------------------------------ Date: 14 Nov 2007 14:58:50 -0500 From: brooks@cuebid.zko.hp.nospam (Rob Brooks) Subject: Re: SYS$GRANT_LICENSE output codes Message-ID: "Tom Linden" writes: > On Wed, 14 Nov 2007 06:38:49 -0800, Rob Brooks wrote > This is what is in my starlet lib, and it is evidently wrong. I don't know PL/I, so I can't comment. I also don't know who supplied the PL/I SDL backend that would have generated that definition. > dcl SYS$GRANT_LICENSE entry ( > character(*), > character(*), > (1:4) fixed binary(31) reference optional, > any reference optional) > returns (fixed binary(31)); > { Backward compatible definition of service. > iflanguage ada bas basic bliss blissf bliss64 dtr for fortran fortv3 lisp > mac macro nparse pas pascal pli sdml tpu uil > ENTRY SYS$GRANT_LICENSE ALIAS $GRANT_LICENSE PARAMETER ( > CHARACTER DESCRIPTOR NAMED PRDNAM IN TYPENAME CHAR_STRING, > CHARACTER DESCRIPTOR NAMED PRDUCER IN TYPENAME CHAR_STRING, > LONGWORD UNSIGNED DIMENSION 4 NAMED CONTXT OUT DEFAULT 0 TYPENAME > CONTEXT, > ANY NAMED ITMLST IN DEFAULT 0 TYPENAME ITEM_LIST_3 > ) RETURNS LONGWORD TYPENAME COND_VALUE; That definition looks pretty good to me; it specifies that the last param is an item list. What's the problem; how do you think it should read? -- Rob Brooks MSL -- Nashua brooks!cuebid.zko.hp.com ------------------------------ Date: 14 Nov 2007 15:19:19 -0600 From: Kilgallen@SpamCop.net (Larry Kilgallen) Subject: Re: SYS$GRANT_LICENSE output codes Message-ID: In article , brooks@cuebid.zko.hp.nospam (Rob Brooks) writes: > "Tom Linden" writes: >> On Wed, 14 Nov 2007 06:38:49 -0800, Rob Brooks wrote > >> This is what is in my starlet lib, and it is evidently wrong. > > I don't know PL/I, so I can't comment. I also don't know who supplied > the PL/I SDL backend that would have generated that definition. > >> dcl SYS$GRANT_LICENSE entry ( >> character(*), >> character(*), >> (1:4) fixed binary(31) reference optional, >> any reference optional) >> returns (fixed binary(31)); >> { Backward compatible definition of service. > >> iflanguage ada bas basic bliss blissf bliss64 dtr for fortran fortv3 lisp >> mac macro nparse pas pascal pli sdml tpu uil >> ENTRY SYS$GRANT_LICENSE ALIAS $GRANT_LICENSE PARAMETER ( >> CHARACTER DESCRIPTOR NAMED PRDNAM IN TYPENAME CHAR_STRING, >> CHARACTER DESCRIPTOR NAMED PRDUCER IN TYPENAME CHAR_STRING, >> LONGWORD UNSIGNED DIMENSION 4 NAMED CONTXT OUT DEFAULT 0 TYPENAME >> CONTEXT, >> ANY NAMED ITMLST IN DEFAULT 0 TYPENAME ITEM_LIST_3 >> ) RETURNS LONGWORD TYPENAME COND_VALUE; > > That definition looks pretty good to me; it specifies that the last param > is an item list. What's the problem; how do you think it should read? The prior complaint (not quoted above) was a lack of strong typing. The SDL quoted says it is not a pointer to an item list, but a pointer to "ANY" which is weak typing anyway you slice it. I find no definition in the SDL for ITEM_LIST_3, for instance the definition of an array of 12 byte entries terminated by any such entry whose first four bytes are zero. $ sear WHEREVER:[*.lis]*.sdl/match=and item_list_3,aggregate %SEARCH-I-NOMATCHES, no strings matched I am not surprised, because the SDL language would have to be enhanced to express that, which I presume was Tom's complaint. ------------------------------ Date: Wed, 14 Nov 2007 21:46:51 GMT From: VAXman- @SendSpamHere.ORG Subject: system constants in COBOL Message-ID: I've been working with a customer on some COBOL code. Bleach! What sorry nonsense. Anyway, they need to invoke some system routines. I written them and even figured out how to create ITEM-LISTS. What I can't seem to figure out is how to implement the constants for system services such as those I would normally find in the Macro library as, for example, $LNMDEF. I've had to use: 01 blah PIC 9(4) COMP VALUE IS EXTERNAL LNM$_STRING. and then create and compile a small macro with: $LNMDEF GLOBAL .END and link with the program. Is there a way to include such literal values in COBOL? -- VAXman- A Bored Certified VMS Kernel Mode Hacker VAXman(at)TMESIS(dot)COM "Well my son, life is like a beanstalk, isn't it?" http://tmesis.com/drat.html ------------------------------ Date: Thu, 15 Nov 2007 11:55:41 +1100 From: Jim Duff Subject: Re: system constants in COBOL Message-ID: <473b9911@dnews.tpgi.com.au> VAXman- @SendSpamHere.ORG wrote: > I've been working with a customer on some COBOL code. Bleach! What sorry > nonsense. > > Anyway, they need to invoke some system routines. I written them and even > figured out how to create ITEM-LISTS. What I can't seem to figure out is > how to implement the constants for system services such as those I would > normally find in the Macro library as, for example, $LNMDEF. I've had to > use: > > 01 blah PIC 9(4) COMP VALUE IS EXTERNAL LNM$_STRING. > > and then create and compile a small macro with: > > $LNMDEF GLOBAL > .END > > and link with the program. > > Is there a way to include such literal values in COBOL? > For some obscure reason best known to the COBOL compiler team, some system constants are resolved and some are not. For the ones that are not, the manual demonstrates how to create a macro just as you have done to resolve them. For example, see: Here's an example showing logical name translation in COBOL. Additionally, it demonstrates that some constants do resolve... $ create lnm.cob $ deck IDENTIFICATION DIVISION. PROGRAM-ID. LNM. ENVIRONMENT DIVISION. CONFIGURATION SECTION. SPECIAL-NAMES. ENVIRONMENT-NAME IS LOGICAL-NAME ENVIRONMENT-VALUE IS LOGICAL-VALUE. DATA DIVISION. WORKING-STORAGE SECTION. 01 TRANSLATION PIC X(64). 01 RETURN-STATUS PIC S9(09) COMP. 88 SYS$_NORMAL VALUE IS EXTERNAL SS$_NORMAL. 88 SYS$_FISH VALUE IS EXTERNAL SS$_FISH. 88 RMS$_FNF VALUE IS EXTERNAL RMS$_FNF. 01 IO$_READVBLK PIC S9(09) COMP VALUE IS EXTERNAL IO$_READVBLK. 01 TT2$M_XON PIC S9(09) COMP VALUE IS EXTERNAL TT2$M_XON. PROCEDURE DIVISION. P0. DISPLAY "SYS$SYSDEVICE" UPON LOGICAL-NAME. ACCEPT TRANSLATION FROM LOGICAL-VALUE ON EXCEPTION DISPLAY "SYS$SYSDEVICE IS NOT DEFINED!" STOP RUN END-ACCEPT. DISPLAY TRANSLATION. END PROGRAM LNM. $ eod $ cobol lnm $ link lnm $ run lnm DSA0: $ show logical sys$sysdevice "SYS$SYSDEVICE" = "DSA0:" (LNM$SYSTEM_TABLE) $ Jim. -- www.eight-cubed.com ------------------------------ Date: Wed, 14 Nov 2007 21:01:23 -0800 (PST) From: Hein RMS van den Heuvel Subject: Re: system constants in COBOL Message-ID: <8a757094-7ea1-4d1c-ac0d-daeffb86e4b1@d50g2000hsf.googlegroups.com> On Nov 14, 4:46 pm, VAXman- @SendSpamHere.ORG wrote: > I've been working with a customer on some COBOL : > how to implement the constants for system services : > and then create and compile a small macro with: > > $LNMDEF GLOBAL > .END > > and link with the program. Yeah,that's what folks do, If it was me, I'd create a DCL / Perl / Macro-defintion tool to turn the MACRO $xxxdef sources from STARLET,MLB into cobol include (.LIB) files and define the constants in the source instead of having the linker resolve them. fwiw, Hein ------------------------------ Date: Thu, 15 Nov 2007 07:36:04 +0100 From: "Gorazd Kikelj" Subject: Re: system constants in COBOL Message-ID: wrote in message news:f1K_i.1548$ID.437@newsfe08.lga... > I've been working with a customer on some COBOL code. Bleach! What sorry > nonsense. > > Anyway, they need to invoke some system routines. I written them and even > figured out how to create ITEM-LISTS. What I can't seem to figure out is > how to implement the constants for system services such as those I would > normally find in the Macro library as, for example, $LNMDEF. I've had to > use: > > 01 blah PIC 9(4) COMP VALUE IS EXTERNAL LNM$_STRING. > > and then create and compile a small macro with: > > $LNMDEF GLOBAL > .END > > and link with the program. > > Is there a way to include such literal values in COBOL? > > -- > VAXman- A Bored Certified VMS Kernel Mode Hacker VAXman(at)TMESIS(dot)COM > > "Well my son, life is like a beanstalk, isn't it?" > > http://tmesis.com/drat.html Hi Brian, here is a short DCL procedure that create a cobol definitions from MLB module. Then include the definition with include statement in working-storage section. Best, Gorazd P1 = macro module name eg. $SSDEF, $IODEF... P2 = /COMPARE - create 88 entries for comparision /VALUES - create 01 entries P3 = macro library with definitions @MAKELIB $SSDEF /VALUES will create cobol external definitions fo $SS_* values Sorry, if wraps. ===== cut here=== $ ty loc$com:makelib.com $ verify :== 'f$verify("No")' $ $ values = 0 $ if p3.eqs."" then p3 = "sys$library:starlet.mlb" $ libr/extr='p1'/outp='p1'.mar 'p3' $ open/read $i$ 'p1'.mar $ open/write $o$ 'p1'.lib $ fc = "01" $ fp = "pic s9(9) comp value external" $ if p2.eqs."/COMPARE" $ then $ write $o$ f$fao("01 !2AS-return pic s9(9) comp.",f$ext(1,2,p1)) $ fc = " 88" $ fp = "value external" $ endif $ if p2.eqs."/VALUES" $ then $ values = 1 $ fp = "picture s9(9) comp value is " $ endif $loop: read/end=endloop $i$ rec $ if f$ext(0,4,rec).eqs."$EQU" $ then $ field = f$edit(f$elem(1," ",rec),"TRIM") $ value = f$edit(f$elem(2," ",rec),"TRIM") $ write sys$output field $ write $o$ f$fao("!AS !28AS !AS !AS.",fc,field,fp,f$elem(values,"/",field+"/"+value)) $ endif $ goto loop $endloop: $ close $i$ $ close $o$ $ verify == f$verif(verify) $ exit === end=== ------------------------------ Date: Wed, 14 Nov 2007 15:24:18 -0500 From: JF Mezei Subject: Re: TCPIP SMTP and/or BIND Resolver bug Message-ID: <5364e$473b5978$cef8887a$17827@TEKSAVVY.COM> FrankS wrote: > On Nov 14, 7:59 am, JF Mezei wrote: >> The expected behaviour would be for the SMTP symbiont to differentiate >> between a "host not found" and a "dns server not responding" and in the >> later case, would requeue the message for delivery. > > What error is the SMTP log file reporting? (usernames and domains modified to prevent spam harvesting) 14-NOV-2007 05:55:11.34: next open file _$11$DQB0:[USERS.JFMEZEI.MAILBOX]0711140 5551483_JFMEZEI-52A00147.TCPIP_BIKE;1(77) Printing debug_level 2, Domains and recipients: Domain: bikeodissey.com.au Recipient address: UsEr@bikeodissey.com.au Domain part: bikeodissey.com.au Local part: UsEr Address Status: Not done, not delivered. (Requeued) Printing debug_level 2, relevant headers: Return-Path: jfmezei@vixenation.ca Date: Wed, 14 Nov 2007 05:55:14 -0500 (EST) Message-Id: <07111405551483_52A00147@vixenation.ca> From: jfmezei@vixenation.ca To: UsEr@bikeodissey.com.au Subject: Re: White fluffy sheep X-VMS-To: SMTP%"User@bikeodissey.com.au" A permanent error makes this mail undeliverable. Must bounce it. %TCPIP-E-SMTP_UNKHST, remote host unknown, bikeodissey.com.au -SYSTEM-F-NOSUCHNODE, remote node is unknown Printing debug_level 2, Domains and recipients: Domain: bikeodissey.com.au Recipient address: UsEr@bikeodissey.com.au Domain part: bikeodissey.com.au Local part: UsEr Address Status: Not done, not delivered. (Requeued) Printing debug_level 2, relevant headers: Return-Path: jfmezei@vixenation.ca Date: Wed, 14 Nov 2007 05:55:14 -0500 (EST) Message-Id: <07111405551483_52A00147@vixenation.ca> From: jfmezei@vixenation.ca To: UsEr@bikeodissey.com.au Subject: Re: White fluffy sheep X-VMS-To: SMTP%"User@bikeodissey.com.au" I also get a postmaster non-delivery notification right away, with the same message: %TCPIP-E-SMTP_UNKHST, remote host unknown, bikeodissey.com.au -SYSTEM-F-NOSUCHNODE, remote node is unknown followed by the original contents. > > What version of OpenVMS and TCP/IP are installed? DEC AXPVMS OPENVMS V8.3 Platform Installed DEC AXPVMS TCPIP V5.6-9 Full LP Installed $ tcpip TCPIP> show version HP TCP/IP Services for OpenVMS Alpha Version V5.6 on an AlphaServer DS10L 466 MHz running OpenVMS V8.3 TCPIP> > > Do you have the latest TCP/IP patches applied? Not sure. I've been repeatedly told that since VMS is rock solid, VMS customers don't need to check for patches every week/month. ------------------------------ Date: Wed, 14 Nov 2007 16:54:39 -0800 (PST) From: FrankS Subject: Re: TCPIP SMTP and/or BIND Resolver bug Message-ID: <5b5d8082-15c8-453e-9c9a-b1a831be78aa@e1g2000hsh.googlegroups.com> On Nov 14, 3:24 pm, JF Mezei wrote: > (usernames and domains modified to prevent spam harvesting) > > From: jfme...@vixenation.ca > To: U...@bikeodissey.com.au > Subject: Re: White fluffy sheep You're worried about spam issues, but I'd be more worried about your personal reputation. "White fluffy sheep"? What type of unnatural acts have you been discussing? :-) :-) > > Do you have the latest TCP/IP patches applied? > > Not sure. I've been repeatedly told that since VMS is rock solid, VMS > customers don't need to check for patches every week/month. Absolutely true. However, most people will check for patches when they're observing unusual behaviour. In the case of Windows that happens daily, for OpenVMS only every once in a blue moon. There were patches issued for TCP/IP v5.6 in June, and again in September. I didn't check the release notes to see if this problem is discussed, but if you haven't applied the latest ECO then that would be my first step. ------------------------------ Date: Wed, 14 Nov 2007 17:11:49 -0800 (PST) From: FrankS Subject: Re: TCPIP SMTP and/or BIND Resolver bug Message-ID: The TCP/IP v5.6 Management guide describes the following logical name which sounds like it controls the behaviour you are observing. TCPIP$SMTP_NO_MX Symbiont Does not use MX records to route mail. Attempts to translate the domain part of each SMTP address into a host name and send the mail directly to that address. If the host name does not translate to an address, the mail is returned. If the host is not available, the mail is queued again. I believe you would want to make sure TCPIP$SMTP_NO_MX is set to FALSE. ------------------------------ Date: Wed, 14 Nov 2007 20:46:08 -0800 From: Ken Fairfield Subject: VIOC: sizing and performance Message-ID: <5q218iFtlgjuU1@mid.individual.net> We have a four-node VMSCluster consisting of two GS1280 "frames", each partitioned into two nodes. For reasons best known to the vendor of the primary application, Cerner Millennium, we're using VIOC rather than XFC file caching. The cluster is not homogeneous in that one node is dedicated to "non-production" versions of the application, one node is the production database server, and two nodes share the production "application server" duties (tier 2 in a 3-tier architecture). That said, the vendor is asking us to set VCC_MAXSIZE <= 256000, whereas we have it set to 512000 on two nodes, and 768000 on a third. Looking at SHOW MEMORY/CACHE/FULL on the three production nodes, I see Read Hit Rates of 58% (database node, 512000 maxsize), 15% on app server node w/512000 maxsize, and 66% on the other app server node w/768000 maxsize (the non-prod node is at 256000 and has a hit rate of 71%). Question: What is the best, or at least a good, indicator of VIOC effectiveness? Is it the read hit rate, or something else? Secondly, what would one expect the effect would be of *reducing* VCC_MAXSIZE by 50%, or even 75% in the one case? I'm concerned that the vendor is making these recommendations without taking into consideration the actual potential performance impacts. Their comment reads like so: "Sets size for closed file cache. 256,000 is the start value. This should be tuned down later after full load conditions are studied." I.e., this seems like a new-installation setting, not one based on actual running history of over 2 years. And note that every change takes a downtime for Autogen then reboot . Thanks for any and all hints, tips, and/or personal stories. ;-) -Ken -- Ken & Ann Fairfield What: Ken dot And dot Ann Where: Gmail dot Com ------------------------------ Date: Wed, 14 Nov 2007 23:43:09 -0600 From: "Forster, Michael" Subject: RE: VIOC: sizing and performance Message-ID: <001401c8274a$6b27ab9e$37146a8d@mcwcorp.net> What are your Cache resident memory section sizings and how much total = and free physical memory? Cache and Oracle systems have their own memory structures and are not = benefitted by XFC. On AIX nodes we would reduce the file cache sizing as = to not take memory away from Cache memory structures. What software is on the tier two? Any Cache server page applications? -----Original Message----- From: "Ken Fairfield" To: "Info-VAX@Mvb.Saic.Com" Sent: 11/14/07 10:49 PM Subject: VIOC: sizing and performance We have a four-node VMSCluster consisting of two GS1280 "frames", each partitioned into two nodes. For reasons best known to the vendor of the primary application, Cerner Millennium, we're using VIOC rather than XFC file caching. The cluster is not homogeneous in that one node is dedicated to "non-production" versions of the application, one node is the production database server, and two nodes share the production "application server" duties (tier 2 in a 3-tier architecture). That said, the vendor is asking us to set VCC_MAXSIZE <=3D 256000, whereas we have it set to 512000 on two nodes, and 768000 on a third. Looking at SHOW MEMORY/CACHE/FULL on the three production nodes, I see Read Hit Rates of 58% (database node, 512000 maxsize), 15% on app server node w/512000 maxsize, and 66% on the other app server node w/768000 maxsize (the non-prod node is at 256000 and has a hit rate of 71%). Question: What is the best, or at least a good, indicator of VIOC effectiveness? Is it the read hit rate, or something else? Secondly, what would one expect the effect would be of *reducing* VCC_MAXSIZE by 50%, or even 75% in the one case? I'm concerned that the vendor is making these recommendations without taking into consideration the actual potential performance impacts. Their comment reads like so: "Sets size for closed file cache. 256,000 is the start value. This should be tuned down later after full load conditions are studied." I.e., this seems like a new-installation setting, not one based on actual running history of over 2 years. And note that every change takes a downtime for Autogen then reboot . Thanks for any and all hints, tips, and/or personal stories. ;-) -Ken --=20 Ken & Ann Fairfield What: Ken dot And dot Ann Where: Gmail dot Com ------------------------------ End of INFO-VAX 2007.625 ************************