0 INFO-VAX	Fri, 13 Feb 2004	Volume 2004 : Issue 87      Contents: Re: Allocation classes and SAN Re: Allocation classes and SAN Re: Day of Year DCL procedure  Re: Day of Year DCL procedure 8 Re: HP/Intel Itanium is dead, long life to HP/AMD AthlonP Re: Intel and Microsoft provide higher 32-bit applications performance on Itaniu: Re: It is almost certain now, INTEL will have 64bit x86 !!: Re: It is almost certain now, INTEL will have 64bit x86 !!: Re: It is almost certain now, INTEL will have 64bit x86 !!: RE: It is almost certain now, INTEL will have 64bit x86 !!% Re: Making images of RL02's under VMS  Re: Microvax vs Vaxstation 3100  Re: Microvax vs Vaxstation 3100 1 Re: OpenGL Drawing, Motif on PseudoColor, ZLXp-E1  Re: OpenVMS Feelings Re: OpenVMS FeelingsD Re: OpenVMS thrashes slowaris and aix in cluster tco/virus study ...K Re: OpenVMS thrashes slowaris and aix in cluster tco/virus study ... again! K Re: OpenVMS thrashes slowaris and aix in cluster tco/virus study ... again! K Re: OpenVMS thrashes slowaris and aix in cluster tco/virus study ... again! K Re: OpenVMS thrashes slowaris and aix in cluster tco/virus study ... again! & OT: Windows has become open source :-)0 php_value modifier in CSWS_PHP 1.2 and CSWS 1.3?4 Re: php_value modifier in CSWS_PHP 1.2 and CSWS 1.3?1 puzzle involving ALLOCLASS and SET VOLUME/REBUILD D Re: QUESTION - Basic setup of SMTP , simple e-mail notification taskD Re: QUESTION - Basic setup of SMTP , simple e-mail notification taskD RE: QUESTION - Basic setup of SMTP , simple e-mail notification taskD Re: QUESTION - Basic setup of SMTP , simple e-mail notification taskD Re: QUESTION - Basic setup of SMTP , simple e-mail notification task Re: Rumours of (CPU) Wars ' Re: Searching for VMS 5.x documentation ' Re: Searching for VMS 5.x documentation ' Re: Searching for VMS 5.x documentation ' Re: Searching for VMS 5.x documentation ' Re: Searching for VMS 5.x documentation ' Re: Searching for VMS 5.x documentation  Re: ShannonKnowsHPC.com  ShannonKnowsHPC.com , Re: Solaris crashes itself when /tmp is full4 Re: Solved: "Known problems with V7.3 security MUP?"> Re: SYS$STARTUP:MMOV$STARTUP.COM -- lacks something (priority) Re: Tape Problem Tape Problem Re: Tape Problem Re: Tape ProblemP Re: TCPIP V5.4: DECnet/IP: remote nodes cannot translate address -> name -> name6 Re: Unknown identifier DECW$WS_QUOTA in rightslist.dat6 Re: Unknown identifier DECW$WS_QUOTA in rightslist.dat
 Re: VMS 7.3-2 	 VMS 7.3-2 
 Re: VMS 7.3-2 
 Re: VMS 7.3-2 
 Re: VMS 7.3-2  Re: VMS and Unicode $ Re: VMS kit needed in Brazil (Fabio)$ Re: VMS kit needed in Brazil (Fabio)  VMS kit needed in Brazil (Fabio)- Re: What kind of memory for AlphaServer 1000? = Re: will these non-DEC disks work with my hobbyist equipment? 9 will these non-DEC disks work with my hobbyist equipment? = Re: will these non-DEC disks work with my hobbyist equipment? = Re: will these non-DEC disks work with my hobbyist equipment? = Re: will these non-DEC disks work with my hobbyist equipment? 1 Re: [OT] Solaris crashes itself when /tmp is full 1 Re: [OT] Solaris crashes itself when /tmp is full 1 Re: [OT] Solaris crashes itself when /tmp is full 1 Re: [OT] Solaris crashes itself when /tmp is full 1 Re: [OT] Solaris crashes itself when /tmp is full 1 Re: [OT] Solaris crashes itself when /tmp is full 1 Re: [OT] Solaris crashes itself when /tmp is full 1 RE: [OT] Solaris crashes itself when /tmp is full 1 Re: [OT] Solaris crashes itself when /tmp is full 1 Re: [OT] Solaris crashes itself when /tmp is full 1 Re: [OT] Solaris crashes itself when /tmp is full 1 Re: [OT] Solaris crashes itself when /tmp is full 1 Re: [OT] Solaris crashes itself when /tmp is full 1 Re: [OT] Solaris crashes itself when /tmp is full 1 Re: [OT] Solaris crashes itself when /tmp is full 1 Re: [OT] Solaris crashes itself when /tmp is full 1 Re: [OT] Solaris crashes itself when /tmp is full 1 Re: [OT] Solaris crashes itself when /tmp is full 1 Re: [OT] Solaris crashes itself when /tmp is full   F ----------------------------------------------------------------------  # Date: Fri, 13 Feb 2004 17:35:43 GMT & From: Jilly <jilly@clarityconnect.com>' Subject: Re: Allocation classes and SAN @ Message-ID: <2978412873728979af6d4a576aaeedbd@news.teranews.com>   Scott Greig wrote:   > Hello all: > , > If I want to cluster some nodes with a SAN2 > as the disk interconnect, is the following true: > B > 1) All disks on the SAN are allocation class 1 (i.e. $1$DGAxxxx)   Yes.  B > 2) All tapes on the SAN are allocation class 2 (i.e. $1$MGAxxxx)   Yes.  E > 3) Each Alpha with local disks that are to be served to the cluster K >     should (must) have its own allocation class, so that device names are  >     unique in the cluster.   Yes.   > ? > My point being, that allocation classes cannot be assigned to ? > SAN storage arrays - the convention is allocation class 1 for  > disks and 2 for tapes? >  > TIA  > Scott   , I advocate the following setup for a cluster  K CI HSx controllers = disk allclass = 255 and work down, tape allclass = 191 
 and work down   F Node 1 - ALLOCLASS=10, non-shared SCSI ports PAC = 11-19, shared SCSI G        - ports PAC = 2-9, if there is a need to MSCP serve CI or shared D        - SCSI disks then ALLOCLASS needs to match the class that is         - being served F        - TAPE_ALLOCLASS will need to meet TMSCP needs if serving tapes  G Node 2 - ALLOCLASS = 20, non-shared SCSI ports PAC = 21-29, shared SCSI         - ports PAC = 2-9   Node 3 - etc. etc.   --  C Jilly - Working from Home in the Chemung River Valley - Waverly, NY H       - jilly@clarityconnect.com                      - Brett Bodine fanH       - Mark.Jilson@hp.com                            - since 1975 or soH       - http://www.jilly.baka.com           - http://www.brettbodine.com   ------------------------------  # Date: Fri, 13 Feb 2004 18:09:57 GMT & From: Jilly <jilly@clarityconnect.com>' Subject: Re: Allocation classes and SAN @ Message-ID: <c52d72e88db80cd38cb93b9c06a6200d@news.teranews.com>   Jilly wrote:   > Scott Greig wrote: > 
 >> Hello all:  >>  - >> If I want to cluster some nodes with a SAN 3 >> as the disk interconnect, is the following true:  >>  C >> 1) All disks on the SAN are allocation class 1 (i.e. $1$DGAxxxx)  >  > Yes. > C >> 2) All tapes on the SAN are allocation class 2 (i.e. $2$MGAxxxx)  >  > Yes. > F >> 3) Each Alpha with local disks that are to be served to the clusterL >>     should (must) have its own allocation class, so that device names are >>     unique in the cluster.  >  > Yes. >  >>  @ >> My point being, that allocation classes cannot be assigned to@ >> SAN storage arrays - the convention is allocation class 1 for >> disks and 2 for tapes?  >>   >> TIA >> Scott > . > I advocate the following setup for a cluster > I > CI HSx controllers = disk allclass = 255 and work down, tape allclass =  > 191 and work down  > G > Node 1 - ALLOCLASS=10, non-shared SCSI ports PAC = 11-19, shared SCSI I >        - ports PAC = 2-9, if there is a need to MSCP serve CI or shared E >        - SCSI disks then ALLOCLASS needs to match the class that is  >        - being served H >        - TAPE_ALLOCLASS will need to meet TMSCP needs if serving tapes > I > Node 2 - ALLOCLASS = 20, non-shared SCSI ports PAC = 21-29, shared SCSI  >        - ports PAC = 2-9 >  > Node 3 - etc. etc. >   L Bad form to follow up my own post but also remember to set the PAQ on the DQ" ports also if they are being used.   --  C Jilly - Working from Home in the Chemung River Valley - Waverly, NY H       - jilly@clarityconnect.com                      - Brett Bodine fanH       - Mark.Jilson@hp.com                            - since 1975 or soH       - http://www.jilly.baka.com           - http://www.brettbodine.com   ------------------------------  % Date: Fri, 13 Feb 2004 09:21:03 -0500 2 From: "Stanley F. Quayle" <squayle@insight.rr.com>& Subject: Re: Day of Year DCL procedure. Message-ID: <402C96FF.21525.A21F375@localhost>  1 On 12 Feb 2004 at 18:52, Barry Treahy, Jr. wrote: . > > $ write sys$output f$cvtime(,,"DAYOFYEAR")- > Well, it certainly doesn't work on VMS 7.2:   F That became official in V7.3-2.  Guy Peleg promised a backport kit to < V7.3.  If you need it to go back farther, bug HP about it...  
 --Stan Quayle  Quayle Consulting Inc.  
 ----------C Stanley F. Quayle, P.E. N8SQ  +1 614-868-1363  Fax: +1 614 868-1671 1 8572 North Spring Ct. NW, Pickerington, OH  43147 = Preferred address:  stan@stanq.com       http://www.stanq.com    ------------------------------  % Date: Fri, 13 Feb 2004 10:00:09 -0500  From: norm.raphael@metso.com& Subject: Re: Day of Year DCL procedureQ Message-ID: <OF5E76FFB8.E41964D0-ON85256E39.00520519-85256E39.00526861@metso.com>   7 $ write sys$output f$cvtime("11-FEB-2004",,"DAYOFYEAR")  42 $ sho sys/noprocC OpenVMS V7.3-1  on node NODEA  13-FEB-2004 09:55:38.16  Uptime  211  04:51:22    I "Stanley F. Quayle" <squayle@insight.rr.com> wrote on 02/13/2004 09:21:03  AM:   3 > On 12 Feb 2004 at 18:52, Barry Treahy, Jr. wrote: 0 > > > $ write sys$output f$cvtime(,,"DAYOFYEAR")/ > > Well, it certainly doesn't work on VMS 7.2:  > G > That became official in V7.3-2.  Guy Peleg promised a backport kit to > > V7.3.  If you need it to go back farther, bug HP about it... >  > --Stan Quayle  > Quayle Consulting Inc. >  > ----------E > Stanley F. Quayle, P.E. N8SQ  +1 614-868-1363  Fax: +1 614 868-1671 3 > 8572 North Spring Ct. NW, Pickerington, OH  43147 ? > Preferred address:  stan@stanq.com       http://www.stanq.com  >  >    ------------------------------  % Date: Thu, 12 Feb 2004 19:31:08 -0600 @ From: "David J. Dachtera" <djesys.nospam@NeOaSrPtAhMlNiOnWk.net>A Subject: Re: HP/Intel Itanium is dead, long life to HP/AMD Athlon 5 Message-ID: <402C28DC.E61AED2@NeOaSrPtAhMlNiOnWk.net>    Keith Parris wrote:  > ` > Didier Morandi <no@spam.com> wrote in message news:<4027d2e5$0$28923$626a14ce@news.free.fr>...S > > The paper sez that HP may cancel all developments on iA64 and go AMD/Athlon 64.  > F > This article is just Yet Another Speculation that Intel will produceH > an x86-48 compatible chip, thus trading roles with AMD and becoming an > AMD clone-maker. > D > As even "The Register" puts it, "Would Intel want to admit that itE > thinks a rival's technology is the best way forward?  We doubt it." 3 > http://www.theregister.co.uk/content/3/35201.html  >  > I doubt it, too.  D Likewise, I dount they'll admit it - verbally. It takes a MIGHTY bigC person to admit a mistake, and I haven't exactly seen any giants in  corporate America lately...   H Actions, as they say, speak louder than words, however. If AMD64 catches, on before Itanic, it's Yamhill or Chapter 9.   --   David J. Dachtera  dba DJE Systems  http://www.djesys.com/  ( Unofficial Affordable OpenVMS Home Page: http://www.djesys.com/vms/soho/    ------------------------------  # Date: Thu, 12 Feb 2004 18:37:13 GMT & From: Rick Jones <foo@bar.baz.invalid>Y Subject: Re: Intel and Microsoft provide higher 32-bit applications performance on Itaniu . Message-ID: <tJPWb.303$FA7.0@news.cpqcorp.net>  P Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> wrote: > Rick Jones wrote: * >> What is the definition of "large" here?  E > ~70% but Solaris 9 isn't required for any SPARC platform so I don't B > expect the gap to close completely until there are new platforms > that only support Solaris 9.  C >> And how many of those 12000 applications have had their ISVs run  >> through that test harness?   G > I have no idea, one of the two most important Apps used by my current F > client has. The other is developed on Solaris so they probably arn't( > so interested in the Binary Guarantee.  D Presumeably the ~70% supported on Solaris 9 would include those that( have run through the test harness right?  @ Anyhow, I guess then we are talking about ~12000 * ~0.7 or ~8400 supported on Solaris 9 then.   --  = denial, anger, bargaining, depression, acceptance, rebirth... C                                      where do you want to be today? F these opinions are mine, all mine; HP might not want them anyway... :)A feel free to post, OR email to raj in cup.hp.com  but NOT BOTH...    ------------------------------  % Date: Fri, 13 Feb 2004 10:25:11 +0000 O From: Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> C Subject: Re: It is almost certain now, INTEL will have 64bit x86 !! 0 Message-ID: <c0i8m7$maq$1@new-usenet.uk.sun.com>   Rick Jones wrote: R > Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> wrote: >  >>Rick Jones wrote:  >>F >>>What is the Sun chapter and verse on positioning/promoting the V20z >>>versus the V2[145]0?  >  > B >>If you wan't SPARC/Solaris because of apps support then used theF >>V210/V240. If you want a large number of 64bit apps now then use the >>V210/V240. >  > ? >>If you want a commodity platform and you are happy with Linux # >>or Solaris x86 then buy the v20z.  >  > C > I thought that Sun was also having the V20Z cerified for Windows?   3 I am only talking about what we are shipping today.  >  > : >>Whats your strategy with the DL380 and any of the slower/ >>but much more expensive Itanium based boxes ?  >  > A > I'll go ask the marketing folks.  As for slower, I suspect that  > depends on your benchmark. >   G Nope slower for almost everything with the exception of FP which as you 6 know doesn't sell enough servers to make a difference.   > A >>last time I looked the rx2600 with Itanium 2 CPU's with similar A >>performance to the 3.2 Ghz Xeons in the DL series cost ~$25,500 A >>as compared with ~$7000 for the DL380. The low voltage Itaniums > >>are of course cheaper ~$14000 but they are also much slower. >  > 7 >>The V210/V240 are the same price of very close to the  >>price of the v20z. >  > @ > But much slower - again, perhaps depending on the benchmark :) >   4 About the same speed as a more expensive LV Itanium.   :-)    regards  Andrew Harrison    ------------------------------  # Date: Thu, 12 Feb 2004 18:30:34 GMT & From: Rick Jones <foo@bar.baz.invalid>C Subject: Re: It is almost certain now, INTEL will have 64bit x86 !! 0 Message-ID: <eDPWb.302$FA7.154@news.cpqcorp.net>  P Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> wrote: > Rick Jones wrote: F >> What is the Sun chapter and verse on positioning/promoting the V20z >> versus the V2[145]0?   B > If you wan't SPARC/Solaris because of apps support then used theF > V210/V240. If you want a large number of 64bit apps now then use the > V210/V240.  ? > If you want a commodity platform and you are happy with Linux # > or Solaris x86 then buy the v20z.   A I thought that Sun was also having the V20Z cerified for Windows?   : > Whats your strategy with the DL380 and any of the slower/ > but much more expensive Itanium based boxes ?   ? I'll go ask the marketing folks.  As for slower, I suspect that  depends on your benchmark.  A > last time I looked the rx2600 with Itanium 2 CPU's with similar A > performance to the 3.2 Ghz Xeons in the DL series cost ~$25,500 A > as compared with ~$7000 for the DL380. The low voltage Itaniums > > are of course cheaper ~$14000 but they are also much slower.  7 > The V210/V240 are the same price of very close to the  > price of the v20z.  > But much slower - again, perhaps depending on the benchmark :)  
 rick jones --  = portable adj, code that compiles under more than one compiler F these opinions are mine, all mine; HP might not want them anyway... :)A feel free to post, OR email to raj in cup.hp.com  but NOT BOTH...    ------------------------------    Date: 13 Feb 2004 07:34:19 -08001 From: keithparris_NOSPAM@yahoo.com (Keith Parris) C Subject: Re: It is almost certain now, INTEL will have 64bit x86 !! = Message-ID: <cf15391e.0402130734.11fa74c1@posting.google.com>   [ JF Mezei <jfmezei.spamnot@istop.com> wrote in message news:<4029B0F9.26F21AB7@istop.com>... L > Of course, the minute Intel announces that the 8086 will have lockstep, itP > will be a very loud and clear messages that IA64 is going down the drain since0 > if Tandem/NSK goes 8086, so will HPUX and VMS.  B The NonStop folks have been very busy bees lately. Besides booting> NonStop on Itanium, they've developed the new NonStop AdvancedF Architecture which no longer requires lockstep in CPUs or special diskA hardware -- they've figured out a way to get the same benefits on D industry-standard hardware.  (Just like VMS has been doing.)  Take aB look, starting at http://h71033.www7.hp.com/object/NSADVARCNW.html   ------------------------------  % Date: Fri, 13 Feb 2004 07:36:01 -0800 # From: "Tom Linden" <tom@kednos.com> C Subject: RE: It is almost certain now, INTEL will have 64bit x86 !! 9 Message-ID: <NDEMLKKEBOIFBMJLCECIOEBGCNAA.tom@kednos.com>      -----Original Message-----:   From: Keith Parris [mailto:keithparris_NOSPAM@yahoo.com])   Sent: Friday, February 13, 2004 7:34 AM    To: Info-VAX@Mvb.Saic.Com E   Subject: Re: It is almost certain now, INTEL will have 64bit x86 !!       8   JF Mezei <jfmezei.spamnot@istop.com> wrote in message '   news:<4029B0F9.26F21AB7@istop.com>... B   > Of course, the minute Intel announces that the 8086 will have    lockstep, itC   > will be a very loud and clear messages that IA64 is going down     the drain since 2   > if Tandem/NSK goes 8086, so will HPUX and VMS.   D   The NonStop folks have been very busy bees lately. Besides booting@   NonStop on Itanium, they've developed the new NonStop AdvancedH   Architecture which no longer requires lockstep in CPUs or special diskC   hardware -- they've figured out a way to get the same benefits on F   industry-standard hardware.  (Just like VMS has been doing.)  Take aD   look, starting at http://h71033.www7.hp.com/object/NSADVARCNW.html  1 You mean, like the way Stratus has been doing it?       --- (   Incoming mail is certified Virus Free.<   Checked by AVG anti-virus system (http://www.grisoft.com).B   Version: 6.0.566 / Virus Database: 357 - Release Date: 1/22/2004    --- & Outgoing mail is certified Virus Free.: Checked by AVG anti-virus system (http://www.grisoft.com).@ Version: 6.0.566 / Virus Database: 357 - Release Date: 1/22/2004   ------------------------------  % Date: Fri, 13 Feb 2004 19:25:51 +0100 + From: "Hans Vlems" <hvlems.nieuw@zonnet.nl> . Subject: Re: Making images of RL02's under VMS: Message-ID: <c0j543$18eagd$1@ID-143435.news.uni-berlin.de>  9 "Dutchman2000" <icelord2006@yahoo.com> schreef in bericht 5 news:cd5Xb.6627$PY.5731@newssvr26.news.prodigy.com...  > Hello- > L > I have several unknown RL02 platters that I would like to convert to ImageE > files that I can use under SIMH and hopefully determine the format.  >  > How can I do this using VMS? > I > I can mount the paks as FOREIGN but I'm not sure what to do after that.  >  > Anyone here know?  >  >   K The most easy way to make RL02 backups is on a VAX 8600/8650 because it has L an RL02 console device. I guess that is not the system you've got access to,	 right :-) K Depends on how the RL02's were created. If they have been used with RSX you  can:  I $ MOUNT/FOR DLA0:                                     ! WILL TELL YOU THE  LABEL ! $ MOUNT/STRUCTURE=1 DLA0: <label> 8 $ BACKUP/IMAGE DLA0: DKA200:[SOMEDIR]RL02BACKUP.BCK/SAVE $ DISMOUNT DLA1:  E It depends a bit on the hardware you've got available and the on-disk H structure that's on the RL02. For RT-11 packs you'd use the VMS EXCHANGE utility.   Hans   PS: of stuur me een email, verwijder .nieuw uit m'n mail adres   ------------------------------  % Date: Thu, 12 Feb 2004 21:04:38 -0500 3 From: "Richard B. Gilbert" <rgilbert88@comcast.net> ( Subject: Re: Microvax vs Vaxstation 31000 Message-ID: <S5ydnR7qp4Y6rbHdRVn-vg@comcast.com>  B As I observed here, not long ago, DEC designed every machine from B scratch!  They never reused a case, power supply, memory board or - anything else if they could design a new one.   I So even if the outside looks similar it's likely to be very different on  I the inside.  As I recall, the resemblance between the Micro VAX 3100 and  H the VAXStation 3100 were only skin deep and not even that if you looked  at the rear panel.   JF Mezei wrote:   ( >Something has puzzled me for some time. > M >I was under the impression that the Microvax 3100 model 30 was the same as a L >Vaxstation 3100 model 30, except for some byte in ROM that said otherwise, ! >and the name glued to the front.  > M >However, looking back at a SOC and on various 31000 web pages, it seems that O >inside, they are quite different, with the Microvax using MS44 memory inserted L >vertically in the middle of the motherboard, while the VAXstation has thoseH >large MS42 memory boards  sandwiched horizontally over the motherboard. > J >Is there any background on why those two would be so different when their >names seem to be so similar ? > I >Did I get bad information or were those two really so different inside ? H >(Heck, I have even read that I am supposed to have 3 serial ports on my >vaxstation !) >    >    ------------------------------    Date: 13 Feb 2004 09:58:37 -08007 From: jones.computer.srv@worldnet.att.net (Daryl Jones) ( Subject: Re: Microvax vs Vaxstation 3100= Message-ID: <8a646952.0402130958.4bb75050@posting.google.com>    Dear JF Mezei:  D After reading the other responses, I think could add some clarity to your confusion.   = The MicroVAX and VAXstation were created using the MicroVAX I > processor. What is the difference? The MicroVAX is the low endC VAXserver originally configured using the Qbus. The VAXstation is a F graphics workstation (VWS). The processors have changed over time with< the CVAX, Rigel, Mariah, and the NVAX chip sets. There was a8 VAXstation 8000, which processor was used? I don't know.   Regards, Daryl Jones       [ JF Mezei <jfmezei.spamnot@istop.com> wrote in message news:<402B3BED.38CD984C@istop.com>... ) > Something has puzzled me for some time.  > N > I was under the impression that the Microvax 3100 model 30 was the same as aM > Vaxstation 3100 model 30, except for some byte in ROM that said otherwise,  " > and the name glued to the front. > N > However, looking back at a SOC and on various 31000 web pages, it seems thatP > inside, they are quite different, with the Microvax using MS44 memory insertedM > vertically in the middle of the motherboard, while the VAXstation has those I > large MS42 memory boards  sandwiched horizontally over the motherboard.  > K > Is there any background on why those two would be so different when their  > names seem to be so similar ?  > J > Did I get bad information or were those two really so different inside ?I > (Heck, I have even read that I am supposed to have 3 serial ports on my  > vaxstation !)    ------------------------------  # Date: Thu, 12 Feb 2004 22:38:02 GMT . From: Mike Stroyan <stroyan@hpstryn.fc.hp.com>: Subject: Re: OpenGL Drawing, Motif on PseudoColor, ZLXp-E1/ Message-ID: <efTWb.343$O9.319@news.cpqcorp.net>   L In comp.graphics.api.opengl Stuart Norris <stuie_norris@yahoo.com.au> wrote:G |I have now gotten to the stage where my program compiles and runs, but  |the colors are not been set.   D |All I get is a black background with a white diagonal line, not theB |color I try to set.  I stole the color stuff out of the gl_ivp.c.   |   glClearIndex(6.);   1 should be   glClearIndex((GLfloat)IndexMap[6]);      |   IndexMap[0] = base_pixel; " should be   glIndexi(IndexMap[3]);  .   It is pretty bold assuming that you can use 2 |static int glxConfig[]={GLX_DOUBLEBUFFER, None }; ... F |   cmap = DefaultColormapOfScreen (DefaultScreenOfDisplay (display)); in combination with E |   vi = glXChooseVisual(display, DefaultScreen(display), glxConfig); = However, you seem to have gotten away with it on your system. ? A more defensive approach would be to use glXGetConfig with the D default visual to see if it supports the GLX_DOUBLEBUFFER attribute.   --  ! Mike Stroyan, mike.stroyan@hp.com    ------------------------------  % Date: Fri, 13 Feb 2004 08:15:56 +0100 " From: Didier Morandi <no@spam.com> Subject: Re: OpenVMS Feelings 4 Message-ID: <402c79b3$0$28768$626a14ce@news.free.fr>   Sue Skonetski wrote:  G > Something like - Most often asked question on the VMS web page is How  > do we reboot?    Sue, I *love* that one... :-)    D. --  7               Discover the FutureVAX: www.futurevax.com 3                       Discover VAXUS: www.vaxus.org   G    didier morandi ~ sarl au capital de 8 000 euros ~ Revendeur agr HP J Expertise en environnement DIGITAL ~ Formation ~ Programmation ~ MigrationI    Offshore ~ 5 av. A. Durand 31700 Blagnac France. Tl: 33(0)5 6131 6287 F      SIRET 448 694 851 00016 RCS Toulouse http://www.didiermorandi.com   ------------------------------    Date: 13 Feb 2004 03:50:17 -0800. From: alexdaniels@themail.co.uk (Alex Daniels) Subject: Re: OpenVMS Feelings = Message-ID: <9f7f13a8.0402130350.4e2dbe73@posting.google.com>   v susan_skonetski@hotmail.com (Sue Skonetski) wrote in message news:<857e9e41.0402121922.75a63c66@posting.google.com>... >  Fight partners G > that don't want to port their application.  Fight reps that forget to H > sell it.  You are the customers, you have the money you make the call.  A Ok well can HP port AMS to VMS, why should I have to run Linux or - TRU64 to monitor my GS boxes, that run VMS ??   D So much for all the UNIX Portability initative in 7.3-2, if we can'tF even have essential tools needed to monitor Alpha's ported by HP, then! I wont even look to ask partners.    ------------------------------    Date: 13 Feb 2004 10:03:12 -0800' From: icerq4a@spray.se (David Svensson) M Subject: Re: OpenVMS thrashes slowaris and aix in cluster tco/virus study ... = Message-ID: <734da31c.0402131003.3adaa74a@posting.google.com>    Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> wrote in message news:<c0dh2b$286$1@new-usenet.uk.sun.com>...  > Wayne Sewell wrote: B > >>>IF YOU DON'T LIKE IT, DON'T READ IT!!!!!!!!!!!!!!!!!!!!!!!!!!& > >>>IT PROMOTES THE MERITS OF VMS ... > >  > >  > > K > > I'm all for that.  It's the sneering at solaris that's the problem.  If Q > > andy-boy was not infesting this newgroup, that would not be an issue.  But he R > > is, so slams at sun just cause another round of useless traffic where the same' > > bullshit is repeated over and over.  > >  > >  > 8 > Except that sadly it isn't BS at least not on my part.   It often is.  	 > Regards  > Andrew Harrison    ------------------------------    Date: 13 Feb 2004 05:37:47 -0800( From: bob@instantwhip.com (Bob Ceculski)T Subject: Re: OpenVMS thrashes slowaris and aix in cluster tco/virus study ... again!< Message-ID: <d7791aa1.0402130537.93329a8@posting.google.com>  v keithparris_NOSPAM@yahoo.com (Keith Parris) wrote in message news:<cf15391e.0402121503.2f0992db@posting.google.com>... > Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> wrote in message news:<c0abfl$qpj$1@new-usenet.uk.sun.com>... > > > The TechFoolish report you refer to was released some time: > > ago, its is complete tosh, it has been comprehensively; > > trashed at least twice but still you see fit to drag it   > > up again for another airing. > F > I guess you must not have actually looked at the document.  If you'd9 > even read the cover page, you'd have seen the new date.  > A > It's a new version, based on fresh interviews.  Methodology has E > changed a bit, as well.  Downtime due to viruses and worms has been  > added for this round, also.  > F > You might be surprised to learn that it even has Sun with the lowestF > TCO under certain circumstances (cases where the cost of downtime isF > very low -- of course, if the cost of downtime is very low, then who8 > needs a cluster for high availability? one might ask). > G > Do look at the availability results.  They are quite compellingly and + > dramatically better for OpenVMS Clusters.  > A > (VMS might have done even better, but TechWise chose to exclude F > downtime due to natural disasters, as they stated the purpose of the, > study did not include disaster tolerance.)  > now don't try to confuse Andrew by asking him to actually read= what he responds to ... he may forget what he learned in suns C marketing response classes ... "we are the best, deny the rest" ...    ------------------------------    Date: 13 Feb 2004 07:43:05 -0800( From: bob@instantwhip.com (Bob Ceculski)T Subject: Re: OpenVMS thrashes slowaris and aix in cluster tco/virus study ... again!= Message-ID: <d7791aa1.0402130743.4cb4c498@posting.google.com>    Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> wrote in message news:<c0i9ip$mmv$1@new-usenet.uk.sun.com>...  > Keith Parris wrote:  > > Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> wrote in message news:<c0abfl$qpj$1@new-usenet.uk.sun.com>...  > > > > >>The TechFoolish report you refer to was released some time: > >>ago, its is complete tosh, it has been comprehensively; > >>trashed at least twice but still you see fit to drag it   > >>up again for another airing. > >  > > H > > I guess you must not have actually looked at the document.  If you'd; > > even read the cover page, you'd have seen the new date.  > > C > > It's a new version, based on fresh interviews.  Methodology has G > > changed a bit, as well.  Downtime due to viruses and worms has been  > > added for this round, also.  > >  > D > Its based on the same methodology, warming over BS doesn't make it > any more credible does it. > - > You need to face a few uncomfortable facts.  > > > OpenVMS doesn't scale to more than 16 CPU's reliably, so the< > likelyhood of the TechFoolish study finding enough OpenVMS9 > customers using fully configured GS320/GS1280's running  > OpenVMS is vanishingly low.  > > > So any TCO comparisions of fully configured GS320's and evenB > GS1280's vs fully configured F15K/p690 ignores that very obvious@ > fact that in reality the fully configured GS's running OpenVMSB > don't exist. Sun and IBM can however provide plenty of customers? > who have fully configured F15K/P690's the customer I work for % > has 10 fully or nearly full F15K's.  > B > You then get to the next obvious issue which is that the bulk of? > the study (if it was that) could only be for GS320's you need > > to have at least 1 years experience of the system before youB > could reliably answer TCO study questions on it and the GS320 is: > 3-4x slower than the F15K and 2-3x slower than the P690. > > > As I said earlier warmed over BS is still BS. Or put another > way BS in BS out.  >  > 	 > Regards  > Andrew Harrison   > so you provide more bs?  The fact of this study is that VMS is? "UNHACKABLE" and slowaris and aix are not ... look at the virus ; numbers ... all you continue to spout is old gs320 platform ? numbers, but you say nothing about OpenVMS itself which is what : the report is really about ... weither VMS is running on a= gs1280 or soon to be itanium, it is the most secure, reliable < platform on the planet ... numbers AND customers don't lie, - only marketing employees ... deal with it ...    ------------------------------    Date: 13 Feb 2004 09:30:59 -0800' From: icerq4a@spray.se (David Svensson) T Subject: Re: OpenVMS thrashes slowaris and aix in cluster tco/virus study ... again!= Message-ID: <734da31c.0402130930.612e0dc0@posting.google.com>    Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> wrote in message news:<c0i9ip$mmv$1@new-usenet.uk.sun.com>...  > Keith Parris wrote:  > > Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> wrote in message news:<c0abfl$qpj$1@new-usenet.uk.sun.com>...  > > > > >>The TechFoolish report you refer to was released some time: > >>ago, its is complete tosh, it has been comprehensively; > >>trashed at least twice but still you see fit to drag it   > >>up again for another airing. > >  > > H > > I guess you must not have actually looked at the document.  If you'd; > > even read the cover page, you'd have seen the new date.  > > C > > It's a new version, based on fresh interviews.  Methodology has G > > changed a bit, as well.  Downtime due to viruses and worms has been  > > added for this round, also.  > >  > D > Its based on the same methodology, warming over BS doesn't make it > any more credible does it. > - > You need to face a few uncomfortable facts.  > 7 > OpenVMS doesn't scale to more than 16 CPU's reliably,   " Do you have any evidence of that ? It is not my experience.G And recent things such as GS1280 with 7.3-1 and 7.3-2 scales very good.n  < > likelyhood of the TechFoolish study finding enough OpenVMS9 > customers using fully configured GS320/GS1280's runningo > OpenVMS is vanishingly low.i  G I have customers who run GS320 fully configured. Not that hard to find.   > > So any TCO comparisions of fully configured GS320's and evenB > GS1280's vs fully configured F15K/p690 ignores that very obvious@ > fact that in reality the fully configured GS's running OpenVMS > don't exist.  7 You clearly don't know what you are talking about here.i Such configurations do exists.   ------------------------------    Date: 13 Feb 2004 09:59:03 -0800' From: icerq4a@spray.se (David Svensson)nT Subject: Re: OpenVMS thrashes slowaris and aix in cluster tco/virus study ... again!< Message-ID: <734da31c.0402130959.bd9e763@posting.google.com>   Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> wrote in message news:<c0i9ip$mmv$1@new-usenet.uk.sun.com>...r > Keith Parris wrote:a > > Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> wrote in message news:<c0abfl$qpj$1@new-usenet.uk.sun.com>...r > > > > >>The TechFoolish report you refer to was released some time: > >>ago, its is complete tosh, it has been comprehensively; > >>trashed at least twice but still you see fit to drag itt  > >>up again for another airing. > >  > > H > > I guess you must not have actually looked at the document.  If you'd; > > even read the cover page, you'd have seen the new date.p > > C > > It's a new version, based on fresh interviews.  Methodology hasMG > > changed a bit, as well.  Downtime due to viruses and worms has beenr > > added for this round, also.  > >  > D > Its based on the same methodology, warming over BS doesn't make it > any more credible does it. >  > B > You then get to the next obvious issue which is that the bulk of? > the study (if it was that) could only be for GS320's you need.> > to have at least 1 years experience of the system before youB > could reliably answer TCO study questions on it and the GS320 is: > 3-4x slower than the F15K and 2-3x slower than the P690.  F Are you saying that you think GS1280's are less reliable that GS320's?  C One of the points in this study is also that the OpenVMS OS is moreiD reliable than other systems. This is not a new thing for most people though.i   ------------------------------  % Date: Fri, 13 Feb 2004 11:37:22 -0500e* From: JF Mezei <jfmezei.spamnot@istop.com>/ Subject: OT: Windows has become open source :-)i) Message-ID: <402CFD3C.11DBC7DC@istop.com>(  I Microsoft has rung the alarm bells, announcing to the world that parts ofCN Windows NT and Windows 2000 source code have been released to the net.  Why isM this important ? Because the media are now specualting that hackers will takeeL apart that code and find even more flaws in Windows, resulting ina  new wave of virus attacks.s  L Perhaps this was done on purpose when Gates realised that his own staff wereG totally ineffective at making Widnows save, robust and secure, so GatesrL unleashed the code on the net to see how quickly all the faults can be found (and then fixed) ....l    G However, this makes for interesting copyright issues. If a company's IT/M infrastruicture is so insecure that its copyrighted materials can flow out ofrF the vaults so easily, can the company really claim a copyright on it ?   ------------------------------  % Date: Fri, 13 Feb 2004 12:11:44 +0000O) From: Witchy <news@sruasonidyranib.co.uk> 9 Subject: php_value modifier in CSWS_PHP 1.2 and CSWS 1.3?O8 Message-ID: <p6fp20p89cjcnl0v2d518v5mar5ptt9qum@4ax.com>  	 Hi folks,i  C Does anyone know if this modifier is supported in CSWS_PHP 1.2? ThefA version of PHP it's based on (4.3.2?) supports it 'cos that's theS3 version I'm running on my existing Linux webserver.l  E Having puzzled for a while I noticed that php_value is even commentedh@ out in PHP.INI which is why I'm asking. The string exists in theF module that Apache loads so perhaps it's a CSWS 1.3 thing and I should% maybe look towards upgrading to 2.0? c! The linux server runs 2.0.47.....   F System setup: Alpha 3000-300LX, hobbyist VMS 7.3 (which is why I can'tC run CSWS 2.0) plus critical patches PCSI, UPDATE v3, SYS and ACRTL.i  B Hopefully it's something I've missed - it'd be uber-cool to run my, museum website on one of its inhabitants :o)   -- cheers,)   witchy/binarydinosaurs   ------------------------------  # Date: Fri, 13 Feb 2004 17:35:44 GMT ) From: "Rick Barry" <richard.barry@hp.com>d= Subject: Re: php_value modifier in CSWS_PHP 1.2 and CSWS 1.3?l/ Message-ID: <QV7Xb.382$G_.304@news.cpqcorp.net>h  6 "Witchy" <news@sruasonidyranib.co.uk> wrote in message2 news:p6fp20p89cjcnl0v2d518v5mar5ptt9qum@4ax.com... > Hi folks,e >oE > Does anyone know if this modifier is supported in CSWS_PHP 1.2? ThecC > version of PHP it's based on (4.3.2?) supports it 'cos that's thel5 > version I'm running on my existing Linux webserver.   J Yes, but it only works with directives that can be modified by PHP_INI_ALL= or PHP_INI_PERDIR (http://us4.php.net/configuration.changes).d  > For example, add the following line to your mod_php.conf file:   PHP_VALUE max_execution_time 60   9 Restart Apache, then look at http://.../php/php_info.php.e  
 Rick Barry OpenVMS Systems Software Group Hewlett-Packard CompanyC
 Nashua, NH   ------------------------------  + Date: Thu, 12 Feb 2004 23:06:44 +0000 (UTC)iP From: helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply): Subject: puzzle involving ALLOCLASS and SET VOLUME/REBUILD$ Message-ID: <c0h0u4$tv7$1@online.de>  K SYSMAN> do pipe sh sys/noproc ; write sys$output f$getsyi("hw_name") ; SET   VOLUME/REBUILD/LOG DISK$SCRATCH-2 %SYSMAN-I-OUTPUT, command execution on node DANEELI OpenVMS V7.3  on node DANEEL  13-FEB-2004 00:01:55.57  Uptime  3 15:37:37c
 VAX 4000-100An) %SET-I-MODIFIED, _ELIJAH$DKA200: modified92 %SYSMAN-I-OUTPUT, command execution on node GLADIAK OpenVMS V7.3-1  on node GLADIA  13-FEB-2004 00:02:02.89  Uptime  3 15:37:27s DEC 3000 - M300LX & %SET-I-MODIFIED, _$44$DKA200: modified2 %SYSMAN-I-OUTPUT, command execution on node ELIJAHI OpenVMS V7.3  on node ELIJAH  13-FEB-2004 00:02:03.78  Uptime  3 15:38:09a VAXstation 4000-60) %SET-I-MODIFIED, _ELIJAH$DKA200: modified  SYSMAN>a  F Why is the disk sometimes referred to with the ALLOCLASS and sometimesG with the NODE name?  Is it a VAX/ALPHA or a 7.3/7.3-1 thing?  (There is B no 7.3-1 VAX, and I never ran 7.3 ALPHA.)  I only have a couple ofC non-shadowed disks; shadowed ones show up as _DSAxxx: on all nodes.v  A The behaviour is the same regardless of what machine the disk is > physically connected to.   ------------------------------  % Date: Fri, 13 Feb 2004 10:20:17 -0500l* From: JF Mezei <jfmezei.spamnot@istop.com>M Subject: Re: QUESTION - Basic setup of SMTP , simple e-mail notification task=) Message-ID: <402CEB2F.762A8E0B@istop.com>=   BillC wrote:G >     the requirement is simple 'E-Mail' notification, possibly with anS > attachment2 >     of an error-log. TCPIP is version 5.1 ECO 4.  C You need to be a bit more precise about what you do not understand.   N If the task is to send a log file to some internet address, then the following will do it:P  Q $ MAIL thefile.log "smtp%""pastry@chocolate.com""" /subject="This is the subject"r   ------------------------------  % Date: Fri, 13 Feb 2004 16:28:22 +0100 * From: Paul Sture <nospam@sture.homeip.net>M Subject: Re: QUESTION - Basic setup of SMTP , simple e-mail notification task / Message-ID: <402CFB26.2A79D39@sture.homeip.net>    BillC wrote: >  > greetings F >     have been thwarted by the online OpenVMS Documentation and wouldF >     like some guidance in setting up TCPIP / SMTP on an AlphaServer. > G >     the requirement is simple 'E-Mail' notification, possibly with an: > attachment2 >     of an error-log. TCPIP is version 5.1 ECO 4. >   E Have you got basic TCP/IP set up already, or do you need instructionst from the start?t   -- D
 Paul Sture   ------------------------------  % Date: Fri, 13 Feb 2004 07:38:28 -08007# From: "Tom Linden" <tom@kednos.com>kM Subject: RE: QUESTION - Basic setup of SMTP , simple e-mail notification taskQ9 Message-ID: <NDEMLKKEBOIFBMJLCECIEEBHCNAA.tom@kednos.com>a  D I don't think he will be happy with 5.1, particularly because of itsA limited spam filtering capabilities, which is why I went to MX5.3c     -----Original Message-----3   From: Paul Sture [mailto:nospam@sture.homeip.net] )   Sent: Friday, February 13, 2004 7:28 AMt   To: Info-VAX@Mvb.Saic.ComeJ   Subject: Re: QUESTION - Basic setup of SMTP , simple e-mail notification   task         BillC wrote:   > 
   > greetingsmH   >     have been thwarted by the online OpenVMS Documentation and wouldH   >     like some guidance in setting up TCPIP / SMTP on an AlphaServer.   > I   >     the requirement is simple 'E-Mail' notification, possibly with an    > attachment4   >     of an error-log. TCPIP is version 5.1 ECO 4.   >    G   Have you got basic TCP/IP set up already, or do you need instructionsh   from the start?i      -- G   Paul Sture      ---o(   Incoming mail is certified Virus Free.<   Checked by AVG anti-virus system (http://www.grisoft.com).B   Version: 6.0.566 / Virus Database: 357 - Release Date: 1/22/2004    ---8& Outgoing mail is certified Virus Free.: Checked by AVG anti-virus system (http://www.grisoft.com).@ Version: 6.0.566 / Virus Database: 357 - Release Date: 1/22/2004   ------------------------------  % Date: Fri, 13 Feb 2004 17:40:17 +0100 * From: Paul Sture <nospam@sture.homeip.net>M Subject: Re: QUESTION - Basic setup of SMTP , simple e-mail notification tasko0 Message-ID: <402D0C01.57DBD3F5@sture.homeip.net>   BillC wrote: > N > hi - TCPIP is indeed setup and functional. i use IP for printing, TELNET for > interactive sessions,eN >     etc. SHOW SERVICE SMTP shows that the service is enabled... but not much > else:  >  > TCPIP> sho service smtp/fult >  > Service: SMTPt/ >                            State:     EnablediI > Port:               25     Protocol:  TCP             Address:  0.0.0.0sL > Inactivity:          5     User_name: TCPIP$SMTP      Process:  TCPIP$SMTPG > Limit:              10     Active:        0           Peak:         0t > 4 > File:         TCPIP$SYSTEM:TCPIP$SMTP_RECV_RUN.COM > Flags:        Listen >  > Socket Opts:  Rcheck Schecka2 >  Receive:            0     Send:               0 > M > Log Opts:     Acpt Actv Dactv Conn Error Exit Logi Logo Mdfy Rjct TimO Addry@ >  File:        SYS$SPECIFIC:[TCPIP$SMTP]TCPIP$SMTP_RECV_RUN.LOG > 
 > Security >  Reject msg:  not defined  >  Accept host: 0.0.0.0J >  Accept netw: 0.0.0.0e >  > tia.   What is the output for M   TCPIP SHOW CONFIHG SMTPe  D (Please disguise any addresses you don't want showing to spammers in your reply.)  C Here is an old set of config commands I found in my email archives:M   $ UCXp SET CONF ENA SERV SMTP- SET CONF SMTP/GATE=ALT="mailhub.mydomain.com"t- SET CONF SMTP/GATE=GEN="mailhub.mydomain.com"i) SET CONF SMTP/ZONE="mailhub.mydomain.com"o) SET CONF SMTP/OPT=(NOEIG,NORELA,TOP_HEAD)e	 STOP MAIL 
 START MAIL --  
 Paul Sture   ------------------------------  % Date: Fri, 13 Feb 2004 17:44:44 +0100'* From: Paul Sture <nospam@sture.homeip.net>M Subject: Re: QUESTION - Basic setup of SMTP , simple e-mail notification taskf/ Message-ID: <402D0D0C.7AC158A@sture.homeip.net>    Paul Sture wrote:t >    Ooops - typo alert.d    > What is the output for >  > TCPIP SHOW CONFIHG SMTP  >   " Should read TCPIP SHOW CONFIG SMTP   -- t
 Paul Sture   ------------------------------    Date: 13 Feb 2004 09:29:50 -08001 From: keithparris_NOSPAM@yahoo.com (Keith Parris) " Subject: Re: Rumours of (CPU) Wars= Message-ID: <cf15391e.0402130929.3b6cfb59@posting.google.com>   d "Bill Todd" <billtodd@metrocast.net> wrote in message news:<7O6dndj2r4ptoLTdRVn-jQ@metrocast.net>...H > What part of the fact that that is a current-implementation limitationN > rather than an architectural limitation do you find difficult to understand,J > Keith?  After all, Alpha implementations haven't always supported a fullH > 64-bit virtual address space, either (I don't even know if the currentM > implementations do, but my early Alpha Architecture Handbook specifies that:D > the minimum virtual address space size supported by any conformingI > implementation is 43 bits):  did/does that mean that Alpha wasn't/isn'tr > really a 64-bit architecture?o  @ Alpha has always supported a flat 64-bit virtual address space. D Implementations which implemented fewer than 64 virtual address bitsF were always carefully specified to sign-extend all addresses so they'dF be upwardly-compatible with future implementations that supported more virtual address bits.u  D In Opteron's case, from a brief look at their architecture manual atR http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/24593.pdf,? it appears they're doing the same thing.  So I stand corrected.l   ------------------------------  % Date: Fri, 13 Feb 2004 10:48:11 +0000 * From: Nic Clews <sendspamhere@[127.0.0.1]>0 Subject: Re: Searching for VMS 5.x documentation' Message-ID: <c0ia23$gdk$1@lore.csc.com>f   Seghers Bruno wrote: > J > Somebody has told me that there is a beautifull explanation of how works& > memory in the VMS 5.x documentation. > M > He things that it's in the "Tuning" book, and it's a comparison between VMS24 > memory mechanism and the management of a warehouse  C That is right there is a description, and it talks about fork liftshB moving pallets into the work area to and from a warehouse. It is aH reasonable introduction, but it doesn't explain global pages, page faultE clusters, the role of MPW settings (high, low, thrash, etc.), and theuH interaction of the page and swap files and the effects of process memory quotas.s  D It quite simplistically dealt with a world with limited memory, whenB things aren't quite so simple. The current performance guide has a0 reasonable introduction but without the analogy.  G http://h71000.www7.hp.com/doc/73final/6491/6491pro_002.html#memory_chapg  G Without getting too complex, the link above which replaces that book isa more accurate.  F Memory management is pretty complex, but the typical reaction today isG to purchase more memory and increase the quotas. I'm not saying that is1B necessarily the right approach, but it quite often has most of the desired effect.V  F Why do you seek the old description? Some good questions after reading< the above, put here, should help further your understanding. -- u? Regards, Nic Clews a.k.a. Mr. CP Charges, CSC Computer Sciencess nclews at csc dot come   ------------------------------  % Date: Fri, 13 Feb 2004 12:53:35 +0100 * From: Paul Sture <nospam@sture.homeip.net>0 Subject: Re: Searching for VMS 5.x documentation0 Message-ID: <402CC8CF.521D7F94@sture.homeip.net>   Rob Brown wrote: > + > On Thu, 12 Feb 2004, Seghers Bruno wrote:  > C > > ...  and it's a comparison between VMS memory mechanism and the0 > > management of a warehouseo > 6 > I never heard of it being in the docs, but how about >   H I do remember it, but my somewhat rusty memory is telling me it also may( have been in either the V3 or V4 docset.  | <http://eisner.encompasserve.org/htbin/dnqindexform?TEXT=R23945136-23948308-dra4%3A%5Bdecuserve_extracts%5Dvms.full%3B1007>?  
 That gives mem  	 ERROR 403    Unable to fopen() database.      --  
 Paul Sture   ------------------------------  % Date: Fri, 13 Feb 2004 14:47:37 +0100.* From: Paul Sture <nospam@sture.homeip.net>0 Subject: Re: Searching for VMS 5.x documentation0 Message-ID: <402CE389.245C7232@sture.homeip.net>   Bob Koehler wrote: > Z > In article <c0h06c$nmb$1@news.worldonline.be>, "Seghers Bruno" <tips@euronet.be> writes:L > > Somebody has told me that there is a beautifull explanation of how works( > > memory in the VMS 5.x documentation. > >nO > > He things that it's in the "Tuning" book, and it's a comparison between VMSs6 > > memory mechanism and the management of a warehouse > I >    Nope.  It's in the VMS 3.6 version of the doc set.  I'd sure like to / >    know who's decision it was to take it out.   A I'm fairly sure it was there from either 3.0 or maybe 3.1 or 3.2.g  G Yes, a shame it went missing. It was also my opinion that the V4 docsetuD dumbed down on the RMS details, although I could be mixing up the V3* documentation with the RMS-11 manual here.   -- g
 Paul Sture   ------------------------------    Date: 13 Feb 2004 06:17:27 -08000 From: keith.cayemberg@conti.de (Keith Cayemberg)0 Subject: Re: Searching for VMS 5.x documentation= Message-ID: <3a65a5c8.0402130617.7890dce5@posting.google.com>-   Hi Seghers,I  4 Most of the Bookreader ODL's for the last Decade are online at Sysworks...t# http://www.sysworks.com.au/odl.htmla   There you will find the... OpenVMS 5.5 ODL - March 1992B http://www.sysworks.com.au/swadm_dat_root/cddoc04mar2/library.html   Cheers!e   Keith Cayemberg ) IBM Business Services - Hannover, Germanyw    ] "Seghers Bruno" <tips@euronet.be> wrote in message news:<c0h06c$nmb$1@news.worldonline.be>...eJ > Somebody has told me that there is a beautifull explanation of how works& > memory in the VMS 5.x documentation. > M > He things that it's in the "Tuning" book, and it's a comparison between VMSS4 > memory mechanism and the management of a warehouse > ; > Is there anybody who have a digital version of this book.a >  > Thanks a lot >  > Seghers Brunoi	 > Belgiump   ------------------------------    Date: 13 Feb 2004 08:29:37 -0600; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler)t0 Subject: Re: Searching for VMS 5.x documentation3 Message-ID: <X07Af6z7T8yt@eisner.encompasserve.org>l  ] In article <402CE389.245C7232@sture.homeip.net>, Paul Sture <nospam@sture.homeip.net> writes:V > C > I'm fairly sure it was there from either 3.0 or maybe 3.1 or 3.2.   F    I don't think so.  I'm fairly sure I recall putting the update intoD    the doc set when I got the 3.6 update and then reading it because8    the pages looked so good when I was doing the insert.   ------------------------------  % Date: Fri, 13 Feb 2004 16:15:13 +0100 * From: Paul Sture <nospam@sture.homeip.net>0 Subject: Re: Searching for VMS 5.x documentation/ Message-ID: <402CF811.21D0100@sture.homeip.net>-   Bob Koehler wrote: > _ > In article <402CE389.245C7232@sture.homeip.net>, Paul Sture <nospam@sture.homeip.net> writes:b > >$E > > I'm fairly sure it was there from either 3.0 or maybe 3.1 or 3.2.u > H >    I don't think so.  I'm fairly sure I recall putting the update intoF >    the doc set when I got the 3.6 update and then reading it because: >    the pages looked so good when I was doing the insert.  G I stand corrected then. I must have got further up the V3.n path than IhA had remembered when working at the customer site where I read it.i   --  
 Paul Sture   ------------------------------  # Date: Fri, 13 Feb 2004 13:22:25 GMT-3 From: wallacethinmintr@eircom.net (Russell Wallace)a  Subject: Re: ShannonKnowsHPC.com. Message-ID: <402ccf01.7604197@news.eircom.net>  . On Fri, 13 Feb 2004 01:50:30 GMT, "Ken Farmer"& <KFarmer@NOSPAM.SpyderByte.com> wrote:  J >Terry posted some intertesting articles the last few days on his website: >  >http://www.shannonknowshpc.com   B Any that you'd particularly recommend? I looked at a few, but theyD read like they were written by a HP marketing droid - all buzzwords,+ no content. I may have missed some, though.-   -- - "Sore wa himitsu desu."h To reply by email, remove0 the small snack from address.<! http://www.esatclear.ie/~rwallaceR   ------------------------------  # Date: Fri, 13 Feb 2004 01:50:30 GMTs2 From: "Ken Farmer" <KFarmer@NOSPAM.SpyderByte.com> Subject: ShannonKnowsHPC.com< Message-ID: <G3WWb.5199$jx3.484554@twister.southeast.rr.com>  I Terry posted some intertesting articles the last few days on his website:o   http://www.shannonknowshpc.com  ( - 2003 HP Retrospective, Part 1 & 2 of 5 - All's REALLY Well at Dell!  7 Terry's been pumping out a few articles a day all week.n  . Don't forget to click on a few Google ads.  :)   KenM   -- Kenneth Farmer <>< OpenVMS.orgm   ------------------------------  % Date: Fri, 13 Feb 2004 08:02:03 +0100d( From: "Rudolf Wingert" <win@fom.fgan.de>5 Subject: Re: Solaris crashes itself when /tmp is fulli: Message-ID: <MCELKPMOKPMNDNKJNIONAELICKAA.win@fom.fgan.de>   Hello,  N IMHO is, that it is better that Solaris crashes as (like OpenVMS) it tells youI the never ending story "pagefile is full trying to continue". Yes I know,mO sometimes it is possible to stop the process, which made the pagefile full, butdN sometimes you need to reboot the system. but the advantage of OpenVMS is, thatP you will get a warning if the pagefile will go full. This warning starts at 60%., Solaris should implement anything like this.   Best regards Rudolf Wingert    ------------------------------  % Date: Fri, 13 Feb 2004 10:29:28 +0000 O From: Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com>9= Subject: Re: Solved: "Known problems with V7.3 security MUP?"J0 Message-ID: <c0i8u8$maq$3@new-usenet.uk.sun.com>   Chris Moore wrote:K > Just a general impression, but it always seems to me that the bulk of (ifhI > not all) the security patches are directly related to some component orh: > other that was ported from Unix or is Windows-compliant. >   > Well the IP stack and its layered apps has been a major source0 of holes and I believe that it comes from Tru64.  9 But thats the world OpenVMS now exists in like it or not,v: with an annual revenue of ~$250 million aquiring code from; other sources rather than writing code from scratch becomesw a necessity.   Regards  Andrew HArrisono > --K > "Senator, let me state for the record, I am not now, nor have I ever beenC > involved with Figure Skating"  >  > M > "Andrew Harrison SUNUK Consultancy" <Andrew_No.Harrison_No@nospamn.sun.com> = > wrote in message news:c0gb63$2mc$1@new-usenet.uk.sun.com...f >  >>jlsue wrote: >>H >>>On Wed, 04 Feb 2004 10:20:45 +0000, Andrew Harrison SUNUK Consultancy1 >>><Andrew_No.Harrison_No@nospamn.sun.com> wrote:i >>>m >>>t >>>n >>>>Larry Kilgallen wrote: >>>> >>>> >>>>>In articleS > L > <63D30D6E10CFD11190A90000F805FE860492B455@lespaul.process.com>, Mike Duffy > <Duffy@process.com> writes:m >  >>>>>s9 >>>>>>In case anyone had been made nervous by the subjectt >>>>>>line, here is the answer.t >>>>>>: >>>>>>Background:  As soon as I put the mandatory security> >>>>>>update (and prerequisite kits) on my DS10L running V7.3,< >>>>>>it has not stayed up for three consecutive days since. >>>>>>& >>>>>>Answer:  It was a power problem. >>>>>- >>>>>-F >>>>>Thanks a lot Mike, now we will be subjected to endless posts fromD >>>>>Andrew Harrison claiming that VMS patches cause power problems. >>>>A >>>>No the patch exists thats enough to keep the OpenVMS securityt5 >>>>BS merchants on their toes trying to downplay it.n >>>> >>>. >>>uI >>>Except that most of the OpenVMS Security folks in here have never madesH >>>claims like what you're trying to debunk.  There have been one or two@ >>>people, but the rest have been much more reasonable about it. >>>  >>; >>Well  Bob, Bob, Keith, Mark and yourself adds up to 5 notu" >>1 or 2 but lets not split hairs. >> >>K >>>There is a claim - that I, personally, can not verify - that  there have I >>>been far fewer security patches for the OpenVMS admins to install thanD >  > for, > G >>>other OSes they have onsite.  I would have to leave it to the adminsiI >>>themselves to verify whether this claim is true in their environments.r >>>iI >>>Most also acknowledge that the networking stack, web server, etc,, may  >  > also > K >>>introduce some vulnerabilities... but we also recognize that these don't @ >>>always open *security* holes, or even result in a DOS attack. >>>n >>A >>Thats obvious and its not the issue, the issue is that when the1@ >>holes do exist either for DOS or worse they don't get reported >>accurately if at all.g >>B >>In addition because some of the holes appear in layered productsC >>the pervading culture tries to pretend that these are not OpenVMSe	 >>issues.o >>	 >>regardso >>Andrew Harrisone >> >  >  >    ------------------------------  # Date: Thu, 12 Feb 2004 18:14:31 GMT & From: jlsue <jefflsxxxz@sbcglobal.net>G Subject: Re: SYS$STARTUP:MMOV$STARTUP.COM -- lacks something (priority)n8 Message-ID: <4jgn20tdgs2baul84i7ge8k4ccr9875hrc@4ax.com>  A On Wed, 11 Feb 2004 21:47:01 -0600 (CST), sms@antinode.org wrote:   D >   And while I'm complaining about SYS$STARTUP:MMOV$SHUTDOWN.COM, IF >recently (at long last) figured out why I've been getting such choppy@ >MPEG playback after system startup until I manually restart the >MMOV$SERVER process.j >:H >   In SYS$STARTUP:MMOV$STARTUP.COM, the "RUN /DETACHED" command lacks aH >"/PRIORITY" qualifier, so if it's run from a typical batch queue (as itF >was on my system), the MMOV$SERVER process inherits a reduced processG >priority, resulting in poor performance.  (And on an AlpSta 200 4/233,oA >it needs all the performance advantages it can scrape together.)   J Well, a 'typical batch queue' would actually be created without a priorityA setting, making it DEFPRI, or priority 4 by default.  If you haveoC /BASE_PRIORITY as anything other than 4, that's an explicit choice.   G While I have setup queues with different base_priority setting, for anyoJ queue into which I submit startups during system boot, I would ensure that. the startup queue would have /base_priority=4.  / No need to modify the startup procedure itself.d   --- jlss0 The preceding message was personal opinion only.6 I do not speak in any authorized capacity for anyone,  and certainly not my employer.- (get rid of the xxxz in my address to e-mail)w   ------------------------------  % Date: Fri, 13 Feb 2004 17:38:15 +0100 - From: "Winfried Bergmann" <dummy@empuron.com>! Subject: Re: Tape ProblemS: Message-ID: <c0iuhp$17epdn$1@ID-170759.news.uni-berlin.de>  I The term "medium offline" means the tape itself (the medium) is "offline"n? (damaged, broken, unreadable). Have you tried a different tape?   ; "Barry" <barrynospam-news@yahoo.com> schrieb im Newsbeitragf9 news:UZ6Xb.2215$t16.1791084@newssvr28.news.prodigy.com... 
 > Experts, > F > I have a tape drive on an AlphaServer ES40 which shows the following status:e >dB > Magtape BILBO2$MKB500:, device type COMPAQ SDT-10000, is online,E > record-oriented device, file-oriented device, available to cluster,jF > error logging is enabled, controller supports compaction (compaction% > disabled),device supports fastskip.5 >9K >      Error count             1    Operations completed            5619132tK >      Owner process          ""    Owner UIC                      [SYSTEM]rK >      Owner process ID 00000000    Dev Prot            S:RWPL,O:RWPL,G:R,WnK >      Reference count         0    Default buffer size                 512eK >      Density              DDS2    Format                        Normal-11a >eJ >    Volume status:  no-unload on dismount, beginning-of-tape, odd parity. >  > D > However, when I try to use the drive I get the following messages: >l$ > %MOUNT-I-OPRQST, medium is offline6 > %MOUNT-I-OPRQST, Please mount device _BILBO2$MKB500: > < > Then nothing happens, even after a tape has been inserted. > 3 > Any suggestions, other than rebooting the system?a >w	 > Thanks.e > -- > barry in indyu >    ------------------------------  # Date: Fri, 13 Feb 2004 16:31:48 GMTh( From: Barry <barrynospam-news@yahoo.com> Subject: Tape Problem ? Message-ID: <UZ6Xb.2215$t16.1791084@newssvr28.news.prodigy.com>i   Experts,  L I have a tape drive on an AlphaServer ES40 which shows the following status:  A Magtape BILBO2$MKB500:, device type COMPAQ SDT-10000, is online, rD record-oriented device, file-oriented device, available to cluster, E error logging is enabled, controller supports compaction (compaction i# disabled),device supports fastskip.f  I      Error count             1    Operations completed            5619132eI      Owner process          ""    Owner UIC                      [SYSTEM]eI      Owner process ID 00000000    Dev Prot            S:RWPL,O:RWPL,G:R,W,I      Reference count         0    Default buffer size                 512 I      Density              DDS2    Format                        Normal-11V  H    Volume status:  no-unload on dismount, beginning-of-tape, odd parity.    B However, when I try to use the drive I get the following messages:  " %MOUNT-I-OPRQST, medium is offline4 %MOUNT-I-OPRQST, Please mount device _BILBO2$MKB500:  : Then nothing happens, even after a tape has been inserted.  1 Any suggestions, other than rebooting the system?o   Thanks.n --  
 barry in indye   ------------------------------    Date: 13 Feb 2004 09:08:26 -07001 From: nothome@spammers.are.scum (Malcolm Dunnett)> Subject: Re: Tape Problemi- Message-ID: <bFOyzp5Kluna@malvm7.mala.bc.ca.>t  j In article <Te7Xb.2219$t16.1793218@newssvr28.news.prodigy.com>, Barry <barrynospam-news@yahoo.com> writes: > Thanks, Winifried, > H > A tape broke in the drive the other day, and I'll bet there's still a 8 > piece in there. It may be time to put in a call to HP. >   I    If this is a DLT drive then the problem will be that due to the brokensH tape the metal leader on the take-up  spool has become unhooked from theE arm that holds it in place and guides it when attaching to the leadero on the tape cartridge.  J    It's not a terribly complicated procedure to reattach this, but it doesI require removing the drive from its case ( to get at the door interlock )dH and fishing around with some needlenose pliers. Depending on exactly howH it's sitting on the spool you may also have to remove the cover from the# take-up spool to get at the leader.   H    If you have the drive under a maintenance agreement it's a lot easier- to just let HP bring you a replacement drive.g  E   Of course, if this isn't a DLT drive then everything I just said ise probably irrelevant :-)i   ------------------------------  # Date: Fri, 13 Feb 2004 16:49:55 GMT ( From: Barry <barrynospam-news@yahoo.com> Subject: Re: Tape Problemt? Message-ID: <Te7Xb.2219$t16.1793218@newssvr28.news.prodigy.com>s   Thanks, Winifried,  F A tape broke in the drive the other day, and I'll bet there's still a 6 piece in there. It may be time to put in a call to HP.   Winfried Bergmann wrote:K > The term "medium offline" means the tape itself (the medium) is "offline"rA > (damaged, broken, unreadable). Have you tried a different tape?r >      -- y
 barry in indy    ------------------------------  % Date: Fri, 13 Feb 2004 09:57:33 +0100i: From: Karl Rohwedder <extern.karl.rohwedder@volkswagen.de>Y Subject: Re: TCPIP V5.4: DECnet/IP: remote nodes cannot translate address -> name -> name ) Message-ID: <c0i3d6$8sb8@doiweb4.b2x.vwg>i   Karl Rohwedder wrote:   9 > I've just noticed a problem with TCPIP V5.4 (on Alpha):  > G > when connecting from the 5.4 system to another remote system (either g > V5.3 or 5.4) viaD > DECnet-over-IP, the remote system is unable to back translate the  > IP-adress to a name,e.g.? > SHO TERMINAL displays IP$10.... instead of nynode.mydomain...s > H > This leads to a lot of trouble, since our proxy entries are no longer 
 > matched. > K > I've logged a call and the HP specialist was able to reproduce the error n > on his system. > H > A CDI$TRACE shows, that the remote system was able to find the name!?!  I HP just confirmed, that this is a DECnet bug in V7.3-2, according to somesW 'enhancements' in the protocol. Perhaps you should stay at DECnet V7.3-1 when upgradingh to VMS V7.3-2.   -- r  + mit freundlichen Gren | with best regardsp  3 Karl Rohwedder          | it-ingteam(at)t-online.delA                          | extern.karl.rohwedder(at)volkswagen.deh   ------------------------------    Date: 13 Feb 2004 01:24:30 -0800. From: martinkirby12@yahoo.co.uk (Martin Kirby)? Subject: Re: Unknown identifier DECW$WS_QUOTA in rightslist.datr< Message-ID: <224291b.0402130124.2464cdfe@posting.google.com>   Gordon,   E It will have been created by PCSI when you installed DECwindows V1.3.u  H It may be better to modify its value rather than deleting it. Otherwise,8 PCSI may be confused when you come to upgrade to V7.3-2.   Martin Kirby  o peter@langstoeger.at (Peter 'EPLAN' LANGSTOEGER) wrote in message news:<newscache$q0pzsh$che1$1@news.sil.at>... f > In article <c0g4ca$5vs6@news.emirates.net.ae>, "Gordon Robb" <gordon@emirates.nospam.net.ae> writes:C > >I had just completed the installation of a new VMS 7.3-1 cluster A > >when I noticed that I had an new identifier in rightslist.dat.  > >n@ > >The identifier is DECW$WS_QUOTA and has a value of %X8001001. > >tE > >The cluster nodes have separate system disks but use a common disk2@ > >for sysuaf and other common data files. The SYS$NODE_nodenameF > >identifier for the first cluster node has a value of %X8001000 withC > >the other nodes having sequential values from %X8001002 upwards.n > >uF > >I have made the usual documentation and internet searches but can'tJ > >find any relevant information as to the source and reason for existence > >of this identifier. > >rH > >I would like to remove and reuse its value (i.e. rename it) as I need? > >to import data, with specific ACLs, from a previous cluster.o > >U+ > >Any help or advice would be appreciated.s >  > SETt >  >   DISPLAY: >  >     /QUOTA > 8 >           /QUOTA=(quota-name=value[,...]) (Alpha only) > J >        Requires SYSPRV (system privilege) privilege or the DECW$WS_QUOTAE >        rights identifier. If using an executive mode device, SYSNAMs2 >        (system name) privilege is also required.   ------------------------------  + Date: Thu, 12 Feb 2004 21:01:53 +0000 (UTC)b6 From: peter@langstoeger.at (Peter 'EPLAN' LANGSTOEGER)? Subject: Re: Unknown identifier DECW$WS_QUOTA in rightslist.dato1 Message-ID: <newscache$q0pzsh$che1$1@news.sil.at>o  d In article <c0g4ca$5vs6@news.emirates.net.ae>, "Gordon Robb" <gordon@emirates.nospam.net.ae> writes:A >I had just completed the installation of a new VMS 7.3-1 clustera? >when I noticed that I had an new identifier in rightslist.dat.  > > >The identifier is DECW$WS_QUOTA and has a value of %X8001001. >nC >The cluster nodes have separate system disks but use a common diski> >for sysuaf and other common data files. The SYS$NODE_nodenameD >identifier for the first cluster node has a value of %X8001000 withA >the other nodes having sequential values from %X8001002 upwards.i > D >I have made the usual documentation and internet searches but can'tH >find any relevant information as to the source and reason for existence >of this identifier. >eF >I would like to remove and reuse its value (i.e. rename it) as I need= >to import data, with specific ACLs, from a previous cluster.h > ) >Any help or advice would be appreciated.r   SETe  	   DISPLAY2  
     /QUOTA  6           /QUOTA=(quota-name=value[,...]) (Alpha only)  H        Requires SYSPRV (system privilege) privilege or the DECW$WS_QUOTAC        rights identifier. If using an executive mode device, SYSNAMr0        (system name) privilege is also required.   --   Peter "EPLAN" LANGSTOEGERs% Network and OpenVMS system specialiste E-mail  peter@langstoeger.atF A-1030 VIENNA  AUSTRIA              I'm not a pessimist, I'm a realist   ------------------------------  % Date: Fri, 13 Feb 2004 11:08:22 +0100e9 From: Jan-Erik =?iso-8859-1?Q?S=F6derholm?= <aaa@aaa.com>v Subject: Re: VMS 7.3-2' Message-ID: <402CA216.AA396C8D@aaa.com>    Mike Naime wrote:o > H > Our App will not be certified for 7.3-2 until sometime in June.  So noK > production clusters at this time.  July, August, and September should seerL > some 100+ OS packs  upgraded to 7.3-2.  I'm hoping that they will have the& > first update patch out by then.  :-)  1 Any special reason you are *hoping* for a patch ?-5 Wouldn't it be better if no patch was needed at all ?i  	 Jan-Erik.u   ------------------------------  # Date: Thu, 12 Feb 2004 18:25:18 GMTS& From: Lee Mah <lytmah@telusplanet.net> Subject: VMS 7.3-2, Message-ID: <iyPWb.36174$QX4.24478@clgrps13>  C Has anyone implemented VMS 7.3-2 into a production environment yet? ( If so, what issues have you encountered?   ------------------------------  % Date: Thu, 12 Feb 2004 12:43:53 -0800a3 From: Alan Frisbie <Usenet01REMOVE@Flying-Disk.com>u Subject: Re: VMS 7.3-2. Message-ID: <402BE589.9060702@Flying-Disk.com>   Lee Mah wrote:E > Has anyone implemented VMS 7.3-2 into a production environment yet?i* > If so, what issues have you encountered?  @ The firmware on our 1000A was version 5.5 and v7.3-2 wanted 5.7,A but 5.7 was more than three years old and was not on the firmwareSB CD.   I had to go find an old firmware CD.   I don't consider this6 a VMS problem -- more of a hardware maintenance issue.  6 So far (4 days), there has only been one real problem:  D Images that were built on 7.2-1 and linked against TCPIP$ICP_SHR.EXE> would ACCVIO on v7.3-2 when TCPIP$htons is called.   RelinkingB on v7.3-2 cured the problem.   I've been too busy working on otherA things to investigate further or report it to support.   In fact,mA it has been so long since I had a support issue that I don't evenl6 remember how to do it any more.   Ain't VMS wonderful?  @ Other than this one thing, all the other problems were caused byA the fact that the client hardly ever has to reboot the system, sonD certain startup things didn't happen.   For example, a non-autostartD queue didn't get started.   It had been added in the two years since= the previous reboot and the startup procedure wasn't updated.   B I haven't done any quantitative performance tests yet, but the XFCB seems to be a big win here.   We don't have much memory (320 MB onB a 1000A), but we're seeing hit rates well over 90%.   I'm going toA order another 512 MB tomorrow, just based on what we're seeing so0E far.    On the down side, the improved performance may kill my budgete  request for a newer Alpha.   :-)   Alan   ------------------------------  % Date: Thu, 12 Feb 2004 17:10:06 -0800 3 From: Alan Frisbie <Usenet01REMOVE@Flying-Disk.com>g Subject: Re: VMS 7.3-2. Message-ID: <402C23EE.1020905@Flying-Disk.com>    Peter 'EPLAN' LANGSTOEGER wrote:f > In article <402BE589.9060702@Flying-Disk.com>, Alan Frisbie <Usenet01REMOVE@Flying-Disk.com> writes:   > > Lee Mah wrote:  I > > > Has anyone implemented VMS 7.3-2 into a production environment yet?3. > > > If so, what issues have you encountered?  D > > The firmware on our 1000A was version 5.5 and v7.3-2 wanted 5.7,E > > but 5.7 was more than three years old and was not on the firmware F > > CD.   I had to go find an old firmware CD.   I don't consider this: > > a VMS problem -- more of a hardware maintenance issue.  0 > Oh, haven't you heard, they are still online ?D > http://ftp.digital.com/pub/DEC/Alpha/firmware/archive/as1000a.html  A Yes, but it was faster to go find the CD than to download it over.; that sorry excuse for an Internet connection we have.   :-( B We're 23,000 feet from the telco CO, so we can't get ADSL.   ThereC is no cable in the neighborhood.   The best we can get is IDSL (DSL-C over an ISDN line).   144kbps max if nobody else in the building is  downloading photos, music, etc.s  A The irony is that three of the fattest fiber pipes in the country ( are just 50 feet from our property line.   Alan   ------------------------------  # Date: Fri, 13 Feb 2004 14:49:56 GMTt4 From: brad@.gateway.2wire.net (Bradford J. Hamilton) Subject: Re: VMS 7.3-2- Message-ID: <nu5Xb.26163$jk2.65952@attbi_s53>r  c In article <OwGe8OwycYuk@eisner.encompasserve.org>, Kilgallen@SpamCop.net (Larry Kilgallen) writes: d !In article <402CA216.AA396C8D@aaa.com>, Jan-Erik =?iso-8859-1?Q?S=F6derholm?= <aaa@aaa.com> writes: !> Mike Naime wrote: !>> J !>> Our App will not be certified for 7.3-2 until sometime in June.  So noM !>> production clusters at this time.  July, August, and September should see N !>> some 100+ OS packs  upgraded to 7.3-2.  I'm hoping that they will have the( !>> first update patch out by then.  :-) !> e4 !> Any special reason you are *hoping* for a patch ?8 !> Wouldn't it be better if no patch was needed at all ? !n: !For those who don't use Command Recall or F$ENVIRONMENT ? !n> !http://eisner.decus.org/DECUServe/DECnotes/VMS/2841.2006.HTML  I I think Mike meant that applying one "UPDATE" ECO is easier than applyinga< numerous "single" ECO's at multiple customer sites. correct?  J __________________________________________________________________________A Bradford J. Hamilton                    "All opinions are my own" K bMradAhamiPltSon-at-coMmcAast.nPeSt     "Lose the MAPS, and replace '-at-' K0                                          with @"   ------------------------------  % Date: Fri, 13 Feb 2004 16:24:43 -0000f* From: "Richard Brodie" <R.Brodie@rl.ac.uk> Subject: Re: VMS and Unicode+ Message-ID: <c0itoe$hts@newton.cc.rl.ac.uk>@  [ "Per Schrder" <per@mimer.se> wrote in message news:c0irbr$dbv$1@yggdrasil.glocalnet.com...   G > In your C program, use the type wchar_t to represent a single Unicode I > character point. This datatype is 4 bytes long on OpenVMS. (Some peopletA > (Java, Windows) got it all wrong and thought 2 bytes would do.)   L That's a bit harsh. They use 2 bytes because that was one of the fundamentalI design principles in the Unicode specification (number 1, as it happens). M Three major revisions of Unicode later, the landscape looks little different.s   ------------------------------    Date: 13 Feb 2004 03:51:56 -0800. From: fabiopenvms@yahoo.com.br (Fabio Cardoso)- Subject: Re: VMS kit needed in Brazil (Fabio) = Message-ID: <f30679fb.0402130351.18305d9a@posting.google.com>6  m "Hans Vlems" <hvlems.nieuw@zonnet.nl> wrote in message news:<c0gns5$172sfd$1@ID-143435.news.uni-berlin.de>... N > A VMS fan in Brazil is looking for a VMS CD. The machine is a Digital ServerK > 5305, the white box equivalent of the Alpha Server 1200. VMS 7.x would bee# > just fine. Fabio, could you help?r >  > Hans Vlems >  > hvlems at zonnet dot nl:  % Yes ! Give him/her my e-mail address:p   fabiocardoso(AT)yahoo(DOT)coms   Regards2   FC   ------------------------------    Date: 13 Feb 2004 03:52:53 -0800. From: fabiopenvms@yahoo.com.br (Fabio Cardoso)- Subject: Re: VMS kit needed in Brazil (Fabio)S= Message-ID: <f30679fb.0402130352.62994898@posting.google.com>u  m "Hans Vlems" <hvlems.nieuw@zonnet.nl> wrote in message news:<c0gns5$172sfd$1@ID-143435.news.uni-berlin.de>...nN > A VMS fan in Brazil is looking for a VMS CD. The machine is a Digital ServerK > 5305, the white box equivalent of the Alpha Server 1200. VMS 7.x would beM# > just fine. Fabio, could you help?v >  > Hans Vlems >  > hvlems at zonnet dot nle  # By the way - Is he/she a Hobbist ? R   fabiocardoso(AT)yahoo(DOT)com    RegardsO   FC   ------------------------------  % Date: Thu, 12 Feb 2004 21:27:39 +0100o+ From: "Hans Vlems" <hvlems.nieuw@zonnet.nl>o) Subject: VMS kit needed in Brazil (Fabio) : Message-ID: <c0gns5$172sfd$1@ID-143435.news.uni-berlin.de>  L A VMS fan in Brazil is looking for a VMS CD. The machine is a Digital ServerI 5305, the white box equivalent of the Alpha Server 1200. VMS 7.x would bee! just fine. Fabio, could you help?y  
 Hans Vlems   hvlems at zonnet dot nlr   ------------------------------  % Date: Fri, 13 Feb 2004 14:03:50 +0000v( From: John Travell <john@travell.uk.net>6 Subject: Re: What kind of memory for AlphaServer 1000?: Message-ID: <c0ilgb$17elrs$1@ID-120847.news.uni-berlin.de>   DAVID TURNER wrote:E  > The AS1000a uses the following > + > X x 36Bit Parity SIMMS (Gold is standard) L > Largest SIMM is 64MB and these must be installed in fours, unless you have* > an AS1000a 4/266 - then you need 5 simms > (1 is for ECC) > # This isn't quite the whole story...oH EV4 does LONGword ECC and needs the 5th SIMM to be installed to provide  enough bits.F EV5 does QUADword ECC and there are enough bits on 4 SIMMS so the 5th  one is not needed.F So, whether it is a 1000 or 1000a, how many SIMMs you need depends on   what CPU you use in the machine.     John Travell* Independent VMS and Alpha Support analyst. john- at - jomatech - dot - comu http://www.jomatech.com/   ------------------------------    Date: 13 Feb 2004 00:23:15 -0800% From: Bart.Zorn@xs4all.nl (Bart Zorn)gF Subject: Re: will these non-DEC disks work with my hobbyist equipment?= Message-ID: <a98cd882.0402130023.65c80adc@posting.google.com>a  F I am using Fujitsu M2952 and M2954 (4 GB) disks without problems in my? VAXstation 4000-90 and DEC 3000-400 systems for many years now.u   HTH,  	 Bart Zorna  ~ "David J. Dachtera" <djesys.nospam@NeOaSrPtAhMlNiOnWk.net> wrote in message news:<402C42A8.C2704517@NeOaSrPtAhMlNiOnWk.net>...1 > Phillip Helbig---remove CLOTHES to reply wrote:o > > B > > NEC D3847 (12/95) 1.6GB 3145 Cyl, 16 Heads, 63 Sect, 1623.1 MB& > > Quantum Fireball TM 32S012  3.2 GB > > Micropolis 4221   2050 MB  > > Fujitsu M2952SYU  2 GB > > L > > I have 7.3-1 ALPHA (DEC 3000/300 LX, DEC 3000/600, ALPHAstation 255/233)J > > and 7.3 ALPHA (VAXstation 4000/60, VAXstation 4000/90, VAX 4000/100A).L > > Not sure what the SCSI controllers are (just the "stock" ones which came9 > > from the factory)---anyway to find this out from DCL?e > > > > > If a particular disk works, will it work as a system disk? > > L > > (In general, I use DEC disks, and if I don't, then generally not for theI > > system disk.  However, one node still has a non-shadowed system disk,eL > > and one which is too small at that (RF35 on a DSSI bus); I could get twoL > > identical Quantum Fireballs as mentioned above, and have an external boxE > > which has room in it, so might consider that for the system disk, ? > > perhaps putting secondary swap and page files on the RF35.)M > > J > > Two years ago, I bought two used SEAGATE SX910800N (9 GB) and set themL > > up as my DISK$USER shadow set.  They have been running continuously everK > > since (except for a few minutes of downtime while I moved them to othergG > > nodes and/or changed the SCSI ID).  Having had good experience withiJ > > non-DEC disks, and having the opportunity to buy the disks above for a9 > > good price, I'm wondering whether I should go for it.n > C > Are these all SCSI disks? If so, I'd expect some success with theuD > Quantum Fireball and possibly also the Fujitsu. The NEC will be anA > experiment, and I couldn't hazard a guess about the Micropolis.i   ------------------------------  + Date: Thu, 12 Feb 2004 23:19:48 +0000 (UTC)hP From: helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply)B Subject: will these non-DEC disks work with my hobbyist equipment?# Message-ID: <c0h1mj$v6$1@online.de>t  > NEC D3847 (12/95) 1.6GB 3145 Cyl, 16 Heads, 63 Sect, 1623.1 MB" Quantum Fireball TM 32S012  3.2 GB Micropolis 4221   2050 MBp Fujitsu M2952SYU  2 GB  H I have 7.3-1 ALPHA (DEC 3000/300 LX, DEC 3000/600, ALPHAstation 255/233)H and 7.3 ALPHA (VAXstation 4000/60, VAXstation 4000/90, VAX 4000/100A).  I Not sure what the SCSI controllers are (just the "stock" ones which came  5 from the factory)---anyway to find this out from DCL?-  : If a particular disk works, will it work as a system disk?  H (In general, I use DEC disks, and if I don't, then generally not for theE system disk.  However, one node still has a non-shadowed system disk, H and one which is too small at that (RF35 on a DSSI bus); I could get twoH identical Quantum Fireballs as mentioned above, and have an external boxA which has room in it, so might consider that for the system disk,c< perhaps putting secondary swap and page files on the RF35.)   F Two years ago, I bought two used SEAGATE SX910800N (9 GB) and set themH up as my DISK$USER shadow set.  They have been running continuously everG since (except for a few minutes of downtime while I moved them to othermC nodes and/or changed the SCSI ID).  Having had good experience with G non-DEC disks, and having the opportunity to buy the disks above for a .5 good price, I'm wondering whether I should go for it.R   ------------------------------    Date: 13 Feb 2004 08:25:48 -0800. From: al5vf03p02@sneakemail.com (William Webb)F Subject: Re: will these non-DEC disks work with my hobbyist equipment?= Message-ID: <d5ce4b06.0402130825.18e2af41@posting.google.com>t  { helbig@astro.multiCLOTHESvax.de (Phillip Helbig---remove CLOTHES to reply) wrote in message news:<c0h1mj$v6$1@online.de>...n@ > NEC D3847 (12/95) 1.6GB 3145 Cyl, 16 Heads, 63 Sect, 1623.1 MB$ > Quantum Fireball TM 32S012  3.2 GB > Micropolis 4221   2050 MB  > Fujitsu M2952SYU  2 GB > J > I have 7.3-1 ALPHA (DEC 3000/300 LX, DEC 3000/600, ALPHAstation 255/233)J > and 7.3 ALPHA (VAXstation 4000/60, VAXstation 4000/90, VAX 4000/100A).  K > Not sure what the SCSI controllers are (just the "stock" ones which came i7 > from the factory)---anyway to find this out from DCL?i > < > If a particular disk works, will it work as a system disk? > J > (In general, I use DEC disks, and if I don't, then generally not for theG > system disk.  However, one node still has a non-shadowed system disk,eJ > and one which is too small at that (RF35 on a DSSI bus); I could get twoJ > identical Quantum Fireballs as mentioned above, and have an external boxC > which has room in it, so might consider that for the system disk,.> > perhaps putting secondary swap and page files on the RF35.)  > H > Two years ago, I bought two used SEAGATE SX910800N (9 GB) and set themJ > up as my DISK$USER shadow set.  They have been running continuously everI > since (except for a few minutes of downtime while I moved them to othertE > nodes and/or changed the SCSI ID).  Having had good experience with I > non-DEC disks, and having the opportunity to buy the disks above for a  7 > good price, I'm wondering whether I should go for it.d  8 I tried mailing you at the naked address but it bounced.  E Don't forget about the 1.06 GB limit FOR SYSTEM DISKS that exists forrC some, but not all, of the VAX systems you have listed. [See FAQ for0 details]  E And I trust that you meant 7.3 VAX unless you've done something trulyj extraordinary.   :^)V   WWWebb' first initial last name at usps dot govn ========================! William W. Webb- EMS Operations, . OpenVMS Systems Support % USPS DSSC Annex - 4730 Hargrove Road  ( Raleigh, NC 27616-2874 919.325.7500x4186 * * * -b   ------------------------------  + Date: Fri, 13 Feb 2004 10:34:51 -0600 (CST)t From: sms@antinode.orgF Subject: Re: will these non-DEC disks work with my hobbyist equipment?) Message-ID: <04021310345103@antinode.org>-  . From: al5vf03p02@sneakemail.com (William Webb)  L > > I have 7.3-1 ALPHA (DEC 3000/300 LX, DEC 3000/600, ALPHAstation 255/233)L > > and 7.3 ALPHA (VAXstation 4000/60, VAXstation 4000/90, VAX 4000/100A).    G > Don't forget about the 1.06 GB limit FOR SYSTEM DISKS that exists foreE > some, but not all, of the VAX systems you have listed. [See FAQ fora
 > details]  H    Forget about the 1.07GB limit for system disks, which affects none of@ the VAX systems you have listed.  (See the VMS FAQ for details.)  : > I tried mailing you at the naked address but it bounced.  
    Same here: C 550  %TCPIP-E-SMTP_COMMANDERR, SMTP command error astro.multivax.det' Wake me (e-mail) if you get that fixed.h  H ------------------------------------------------------------------------  4    Steven M. Schweda               (+1) 651-699-98183    382 South Warwick Street        sms@antinode-orgr    Saint Paul  MN  55105-2547    ------------------------------  % Date: Fri, 13 Feb 2004 19:52:20 +0100.+ From: "Hans Vlems" <hvlems.nieuw@zonnet.nl>nF Subject: Re: will these non-DEC disks work with my hobbyist equipment?: Message-ID: <c0j6lg$181n6k$1@ID-143435.news.uni-berlin.de>  L "Phillip Helbig---remove CLOTHES to reply" <helbig@astro.multiCLOTHESvax.de>0 schreef in bericht news:c0h1mj$v6$1@online.de...@ > NEC D3847 (12/95) 1.6GB 3145 Cyl, 16 Heads, 63 Sect, 1623.1 MB$ > Quantum Fireball TM 32S012  3.2 GB > Micropolis 4221   2050 MBb > Fujitsu M2952SYU  2 GB >nJ > I have 7.3-1 ALPHA (DEC 3000/300 LX, DEC 3000/600, ALPHAstation 255/233)H > and 7.3 ALPHA (VAXstation 4000/60, VAXstation 4000/90, VAX 4000/100A).J > Not sure what the SCSI controllers are (just the "stock" ones which came7 > from the factory)---anyway to find this out from DCL?o >d< > If a particular disk works, will it work as a system disk? > J > (In general, I use DEC disks, and if I don't, then generally not for theG > system disk.  However, one node still has a non-shadowed system disk, J > and one which is too small at that (RF35 on a DSSI bus); I could get twoJ > identical Quantum Fireballs as mentioned above, and have an external boxC > which has room in it, so might consider that for the system disk,.= > perhaps putting secondary swap and page files on the RF35.)i >eH > Two years ago, I bought two used SEAGATE SX910800N (9 GB) and set themJ > up as my DISK$USER shadow set.  They have been running continuously everI > since (except for a few minutes of downtime while I moved them to other E > nodes and/or changed the SCSI ID).  Having had good experience withuH > non-DEC disks, and having the opportunity to buy the disks above for a7 > good price, I'm wondering whether I should go for it.l >o  J Some (all ?) of the DEC RZnn drives were re-badged third party devices. In front of me sit two drives:g   RZ26L  made by Quantum= RZ28C-E made by Seagate, actually an ST32550N Barracuda driven  G Your reluctance to use non-DEC gear for system disks is understandable.uL Especially if you've been confronted with early SCSI support in VAX/VMS (5.3I ?) But 7.3 (AXP and VAX) is quite tolerant. On the Alpha's I use 10 k rpmpK Seagate Cheetah's for VMS and Tru64 and an ST32171WC for VMS. The 5305 (VMSr 7.3) also uses:i  F Disk $5$DKA0: (OSMIUM), device type QUANTUM ATLAS_V__9_WLS, is online, mounted,I Disk $5$DKA500: (OSMIUM), device type DEC RRD47, is online, file-oriented I Disk $5$DKB0: (OSMIUM), device type COMPAQ ST34501WC, is online, mounted,o file-nK Disk $5$DKB100: (OSMIUM), device type SEAGATE ST32171W, is online, mounted,qJ Disk $5$DKB200: (OSMIUM), device type QUANTUM VIKING II 4.5SCA, is online,J Disk $5$DKB300: (OSMIUM), device type QUANTUM VIKING II 4.5SCA, is online,K Disk $5$DKC0: (OSMIUM), device type DEC RZ1DF-CB, is online, mounted, file-eG Disk $5$DKC100: (OSMIUM), device type DEC RZ2DC-PA, is online, mounted,i file-nG Disk $5$DKC200: (OSMIUM), device type DEC RZ2DC-PA, is online, mounted,c file-   G I find myself using Seagate for SCSI drives and Fujitsu for IDE drives.s   Hans   ------------------------------  % Date: Fri, 13 Feb 2004 08:11:46 +0000t From: Roy Omond <Roy@Omond.net>:: Subject: Re: [OT] Solaris crashes itself when /tmp is full4 Message-ID: <c0i0s3$6o0$1$8300dec7@news.demon.co.uk>  ( Andrew Harrison SUNUK Consultancy wrote: > Fred Kleinsorge wrote: > ' >> "Andrew Harrison SUNUK Consultancy"  * >> <Andrew_No.Harrison_No@nospamn.sun.com>> >> wrote in message news:c0g8os$1qs$1@new-usenet.uk.sun.com... >>C >>> It looks like you have a user/process problem. Does OpenVMS runf+ >>> indefinitely if you fill its disks up ?> >>I >> To some degree, yes.  You might end up in situations where some things K >> (applications) will hang or die as they fail to get space.  In addition,tK >> there are not a lot of situations that would allow a unprivleged user toh" >> trivially fill the system disk. > B > Solaris isn't allowing an unpriviledged user to fill up a system. > disk /tmp isn't generally a disk its memory.  1 Well, looks like this stirred things up a bit :-)t  1 In addition to others' comments, I'd have to say:l  < a) this looks like it's unique to Solaris.  I've not managed> Solaris systems in the last 10 years - and I've not seen this,) ahem, "feature" on any other Unix system.>  @ b) I've never seen VMS crash because any disk (even system disk)8 filled up (and that's in 22 years managing VMS systems).  < c) this Solaris, ahem, "feature" has got to be the stupidest> design I've seen, possibly *ever* in the 35 years I've been in5 this industry (and I've seen plenty of stupid stuff).i  @ d) I understand that it's easy to change this behaviour, but for@ it be like that by default is a serious security hole (yep, here? we go again ;-).  It allows unprivileged users to, by design orp? by accident, crash a system.  At best, this is a classic Denial>3 of Service; at worst, it could lead to catastrophe.t  < I'm flabbergasted because I refused to believe it, even when; told by my Solaris colleagues that it is so.  Looks like mya@ faith in Solaris was unfounded - I shall not go near any Solaris system ever again in my life.h   Codswalloped :-)  	 Roy Omond  Blue Bubble Ltd.   ------------------------------  # Date: Fri, 13 Feb 2004 09:25:22 GMTe0 From: glen herrmannsfeldt <gah@ugcs.caltech.edu>: Subject: Re: [OT] Solaris crashes itself when /tmp is full- Message-ID: <6K0Xb.22699$yE5.98227@attbi_s54>D   Roy Omond wrote:   (snip)  F > 1) From "Feb 2 12:35:06" in the messages.0 log there were error msgs: > reporting that /tmp was full, which is a very bad error: > H > xxxxx unix: WARNING: /tmp: File system full, swap space limit exceeded > H > 2) The OS eventually crashed itself to resolve the problem (this would@ > have cleaned out /tmp) at 16.59 and then came back up at 17.07  ? The common /tmp implementation on Solaris is in virtual memory. . That is why it says swap space limit exceeded.  A While a full disk may cause unix to crash, filling virtual memorye? is more likely to do it.  Once, as part of a test for something2 else did the equivalent of:        for(;;) malloc(1000);    and crashed the system.    Don't do that.   -- glen    ------------------------------  % Date: Fri, 13 Feb 2004 09:59:18 +0000iO From: Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com>n: Subject: Re: [OT] Solaris crashes itself when /tmp is full0 Message-ID: <c0i75m$lmf$1@new-usenet.uk.sun.com>   Bill Gunshannon wrote:2 > In article <402BA1B8.18896617@blueyonder.co.uk>,9 > 	Tim Llewellyn <tim.llewellyn@blueyonder.co.uk> writes:r >  >> >>Roy Omond wrote: >> s >> >>>Any comments ?  >>>e >>>A flabbergasted Roy Omond >>>Blue Bubble Ltd.z >>U >>Sure, it is a unix "weakness". Some syadmins have cron jobs to monitor and clean ups >>/tmp.l >  >  > Here we go again.- > H > No, it is not a Unix weakness.  None of the numerous servers I run nowI > will crash if /tmp fills up and none that I remember in the past (which1H > goes back a couple of decades).  The worst that is likely to happen isE > that some application (vi comes immediately to mind) might not work  > properly.i > E > I am curious enough about this that I plan to set up a machine withaF > Solaris 9 and test this myself.  Even if it turns out to be true, itF > is easily remedied and probably would have been, before the fact, by' > any really experienced Unix Sysadmin.B >   A Sun supplies SunMC with Solaris, one of the agents is a diskspace B monitor, setting a threshold on /tmp and then finding and deleting4 all files over a certain age is simple to impliment.  @ As I said earlier this seems to be a people and process problem.   Regardsr Andrew Harrison  > bill >    ------------------------------    Date: 13 Feb 2004 07:21:02 -0600; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) : Subject: Re: [OT] Solaris crashes itself when /tmp is full3 Message-ID: <UiGusAt+zJVs@eisner.encompasserve.org>    In article <c0i7eh$lua$1@new-usenet.uk.sun.com>, Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> writes: > = > File up the partition that /tmp resides in for example / onI< > a system without this and you will get the same behaviour.  H    So you are sure Solaris will crash if /tmp is an ordinary file system    and you fill it up?  Sad.  C >> b) I've never seen VMS crash because any disk (even system disk)h; >> filled up (and that's in 22 years managing VMS systems).- >> - >  > But other people have.      Who?   ? >> c) this Solaris, ahem, "feature" has got to be the stupidestcA >> design I've seen, possibly *ever* in the 35 years I've been ins8 >> this industry (and I've seen plenty of stupid stuff). >> r > = > Sorry but thats rubbish and if you really don't want to get:= > the performance hike for using tmpfs then you are perfectlye' > free to mount /tmp wherever you like.c  A    Any design that allows an ordinary unpriviledged user to crashoD    the system is a design with a security hole.  Trivial DOS attack.   ------------------------------   Date: 13 Feb 2004 13:46:54 GMT( From: bill@cs.uofs.edu (Bill Gunshannon): Subject: Re: [OT] Solaris crashes itself when /tmp is full: Message-ID: <c0ikge$17ir3o$1@ID-135708.news.uni-berlin.de>  3 In article <nvtT+wmM4iqA@eisner.encompasserve.org>, > 	koehler@eisner.nospam.encompasserve.org (Bob Koehler) writes:k > In article <c0h6ic$16mdfb$1@ID-135708.news.uni-berlin.de>, bill@gw5.cs.uofs.edu (Bill Gunshannon) writes:r >> uJ >> C'mon Bob.  I expect better of you.  The problem with Solaris is uniqueI >> in that it isn't really about filling up /tmp as much as it appears tohG >> be forcing an exhaustion of swap because /tmp is actually in memory.a > = >    Both have the underlying cause of exhausting disk space.-  E But the difference is in exhausting disk space required by the systemmF to run (swap) and exhausting space on a partition that may effect whatJ users can do but not the whole system.  Ages ago sysadmins started linkingH /tmp to something (usually /usr/tmp or /var/tmp) off the root partition.F Today, most sysadmins (and FreebSD by default) put /tmp on a partition9 all by itself assuring minimal impact from filling it up.    >  >>  InI >> have been working with Unix systems of pretty much every flavor since aI >> around 1981 amd have never seen a system crash because /tmp filled up.eG >> I have seen them become unusable, but never crash.                  n > , >    Your experience is different from mine.  F I would love to hear what particualr implementations of Unix you foundE would crash if /tmp filled up, but I assume they were a long time ago D and the chances of getting enough information about these systems toC actually determine what happened is probably nil.  There is nothingrD inherent in just filling up an ordinary partition under Unix that isA likely to crash the kernel. (At least nothing I have ever found.)b   >  uJ >> BSD has had numerous process resource quotas for many years now.  WhichG >> just goes to show how stupid the rush to SYSV and the abandonment of-@ >> commercialization of the version previously known as BSD was. > ; >    BSD was a hell of a lot easier to manage in many ways.T >   G It had more advantages than one can even begin to list.  It still does.aH Over both SYSV and Linux.  Sad that people (and companies) can't seem to	 see that.t   bill   --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>   -   ------------------------------   Date: 13 Feb 2004 13:58:17 GMT( From: bill@cs.uofs.edu (Bill Gunshannon): Subject: Re: [OT] Solaris crashes itself when /tmp is full: Message-ID: <c0il5p$17ir3o$2@ID-135708.news.uni-berlin.de>  0 In article <c0i7eh$lua$1@new-usenet.uk.sun.com>,R 	Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> writes: > Roy Omond wrote:+ >> Andrew Harrison SUNUK Consultancy wrote:  >> v >>> Fred Kleinsorge wrote: >>>s) >>>> "Andrew Harrison SUNUK Consultancy" P, >>>> <Andrew_No.Harrison_No@nospamn.sun.com>@ >>>> wrote in message news:c0g8os$1qs$1@new-usenet.uk.sun.com... >>>>E >>>>> It looks like you have a user/process problem. Does OpenVMS rune- >>>>> indefinitely if you fill its disks up ?o >>>> >>>>K >>>> To some degree, yes.  You might end up in situations where some things M >>>> (applications) will hang or die as they fail to get space.  In addition, M >>>> there are not a lot of situations that would allow a unprivleged user tos$ >>>> trivially fill the system disk. >>>S >>>-D >>> Solaris isn't allowing an unpriviledged user to fill up a system0 >>> disk /tmp isn't generally a disk its memory. >>   >>  4 >> Well, looks like this stirred things up a bit :-) >> -4 >> In addition to others' comments, I'd have to say: >> 0? >> a) this looks like it's unique to Solaris.  I've not managedkA >> Solaris systems in the last 10 years - and I've not seen this,9, >> ahem, "feature" on any other Unix system. >> n > ; > Its a feature of all UNIX systems, the difference is thatd; > Solaris allows you to create a memory resident /tmp whichn@ > can be a big performance win provided you manage it correctly.  A Sorry, I really can't see this.  First, I can't think of anythingND that goes into /tmp that would benefit from it.  Second, considering@ the potential downside as presented here.....  And lastly, if itC worked at all, it would only work for very small or very few files.e@ Once it starts swapping this memory in and out it becoems a lossA rather than a gain.  Unless your telling us it gets memory lockedt- which seems unlikely considering the results.1   > = > File up the partition that /tmp resides in for example / on < > a system without this and you will get the same behaviour.  C No, you won't.  Especially on a well administered system where /tmpoC is it's own partition (the current BSD default).  But even when its8C not, filling up a partition will not crash the kernel.  It may maket; creating files difficult :-) but the system will still run.    bill u   --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>   s   ------------------------------   Date: 13 Feb 2004 13:59:48 GMT( From: bill@cs.uofs.edu (Bill Gunshannon): Subject: Re: [OT] Solaris crashes itself when /tmp is full: Message-ID: <c0il8k$17ir3o$3@ID-135708.news.uni-berlin.de>  3 In article <UiGusAt+zJVs@eisner.encompasserve.org>,)> 	koehler@eisner.nospam.encompasserve.org (Bob Koehler) writes: > In article <c0i7eh$lua$1@new-usenet.uk.sun.com>, Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> writes: >> W> >> File up the partition that /tmp resides in for example / on= >> a system without this and you will get the same behaviour.e > J >    So you are sure Solaris will crash if /tmp is an ordinary file system >    and you fill it up?  Sad.  # I doubt even Solaris would do that.d   bill   -- iJ Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>       ------------------------------  % Date: Fri, 13 Feb 2004 08:06:45 -0600h/ From: "Stuart, Ed" <Ed.Stuart@austinenergy.com> : Subject: RE: [OT] Solaris crashes itself when /tmp is fullT Message-ID: <92EFB80E551BD511B39500D0B7B0CDCC120D2293@ohms.electric.ci.austin.tx.us>  J What about if the disk where the OpenVMS security audit file resides fills up?e   EdE **Please apply a generous amount of all the usual disclaimers here.**h     > -----Original Message-----3 > From: bill@cs.uofs.edu [mailto:bill@cs.uofs.edu]  ) > Sent: Friday, February 13, 2004 8:00 AMA > To: Info-VAX@Mvb.Saic.Com < > Subject: Re: [OT] Solaris crashes itself when /tmp is full >  > 5 > In article <UiGusAt+zJVs@eisner.encompasserve.org>,r@ > 	koehler@eisner.nospam.encompasserve.org (Bob Koehler) writes:< > > In article <c0i7eh$lua$1@new-usenet.uk.sun.com>, Andrew  > Harrison SUNUK t? > > Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> writes:, > >> r> > >> File up the partition that /tmp resides in for example /  > on a system 6 > >> without this and you will get the same behaviour. > > 8 > >    So you are sure Solaris will crash if /tmp is an  > ordinary file system  > >    and you fill it up?  Sad. > % > I doubt even Solaris would do that.  >  > bill >  > -- u@ > Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.   > Three wolvesF > bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. > University of Scranton   |C > Scranton, Pennsylvania   |         #include <std.disclaimer.h>   e >    ------------------------------   Date: 12 Feb 2004 18:42:19 GMT, From: bill@gw5.cs.uofs.edu (Bill Gunshannon): Subject: Re: [OT] Solaris crashes itself when /tmp is full: Message-ID: <c0gheb$16ethc$1@ID-135708.news.uni-berlin.de>  0 In article <c0gdqq$3h5$2@new-usenet.uk.sun.com>,R 	Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> writes: > B > Solaris isn't allowing an unpriviledged user to fill up a system. > disk /tmp isn't generally a disk its memory.  C Sorry Andrew, but in my opinion considering what /tmp gets used for E that is even stupider than putting /tmp on the root (or any critical) 
 partition.   bill   --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>   f   ------------------------------    Date: 12 Feb 2004 15:58:56 -0600; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler)k: Subject: Re: [OT] Solaris crashes itself when /tmp is full3 Message-ID: <gYI4TNsT6bx5@eisner.encompasserve.org>   V In article <c0g5bi$cg0$1$8302bc10@news.demon.co.uk>, Roy Omond <Roy@Omond.net> writes: > E > I have asked my colleagues who deal with Solaris if it is true thathD > Solaris will crash itself if /tmp gets full (for non-Unix persons,C > /tmp is usually a scratch directory where everyone can write to).yB > Astonishingly it is indeed so.  So any black-hat, non-privilegedC > JoeBloggs user could fill /tmp and, *bang*, away goes the system.t+ > Or even a white-hatted user by accident ?e  F    Sigh.  'tis true.  Solaris is likely not the only eunich with this D    problem.  We used to run a cron job to keep /tmp clean overnight,H    as well as cleaning up unreferenced formated man pages, syslog, wtmp,    ...  A    And there are many other things "any black-hat, non-privileged G    JoeBloggs user could" do to hose things up, generally realted to not /    having a concept of process resource quotas.k   ------------------------------    Date: 12 Feb 2004 16:10:52 -0600; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler) : Subject: Re: [OT] Solaris crashes itself when /tmp is full3 Message-ID: <tv24ogAEN7iI@eisner.encompasserve.org>i   In article <c0g8os$1qs$1@new-usenet.uk.sun.com>, Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> writes:  A > It looks like you have a user/process problem. Does OpenVMS runt) > indefinitely if you fill its disks up ?y      It does not crash.s  F    What it does do depends on how it's managed.  If the disk with the H    security audit log fills up the default action is to complain on the J    operator's console and stop processing (appears to hang to most users).D    The system manager can determine where the audit file is and what+    action to take when it's disk gets full.   ?    If the system or a data disk fills up but the security audith?    system does not block processing then VMS will keep running.   G    It will also complain on the operator's console if it can't write to F    the error log or the opertor's log for any reason, including a full    disk.  D    I've seen VMS as old as 2.5 keep on running with it's system diskA    removed.  And VMS as old as 4.0 fully recover without further oD    intervention when it's system disk was removed and then put back.   ------------------------------  # Date: Fri, 13 Feb 2004 02:10:14 GMTl' From: nospam <x@wedontwantyourspam.com>a: Subject: Re: [OT] Solaris crashes itself when /tmp is full5 Message-ID: <BC527D40.2305E%x@wedontwantyourspam.com>e  F in article P8PWb.294494$xy6.1449900@attbi_s02, Bradford J. Hamilton at2 brad@.gateway.2wire.net wrote on 13/02/2004 04:58:  H > In article <c0gb9t$2mc$2@new-usenet.uk.sun.com>, Andrew Harrison SUNUK= > Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> writes: " > !david20@alpha2.mdx.ac.uk wrote: > !snip! > ! E > !As you could for /tmp in fact you could set TMPDIR for each of thei: > !users to a "userhomedir"/tmp directoy if you wanted to. Yuk ;)   > !d > N > I used to make sure that /tmp was on a different disk from / on my Linux and > FreeBSD boxes at home.  C Good practice!  The default place for /tmp is out of / or root/bootoG filesystem.  Root  is usually quite small so having /tmp there is a bitrJ limiting. Though the real reason for a separate file system for /tmp datesH back to some security problems. Having a user writable directory in rootJ (/tmp) once allowed malicious students to make hard links to system files.K Then other try to corrupt them eg core dumps or using names that are likely L to be written too. Also I think you could read protected files with lpd, itsL been a long time since any of this was true so its kind of sketchy  now that I think about it decades later.tH  I assume solaris must be running in some over commit mode with swap andG that the admin has curiously allocated a too large a space to /tmp on aUL memory filesystem with regard to the system usage. Once an OS in over commitF mode fill both swap and memory but still needs to fault a page out andG allocate swap then its faced with few choices. VMS just hangs, Tru64 an K others kill a process and since this allocation was probably the filesystem H driver I might be convinced that reboot was a reasonable choice to make.F   It sounds more like mismanaged than Suns' fault, sorry find a betterJ topic. I don't use solaris and have my own pet peeves with them but if youI going to start this kind of attack, I can think of several things equallye and more bad to say for VMS.	   Mark;)     ------------------------------  % Date: Fri, 13 Feb 2004 15:22:19 +0100d* From: Paul Sture <nospam@sture.homeip.net>: Subject: Re: [OT] Solaris crashes itself when /tmp is full0 Message-ID: <402CEBAB.2AD40B83@sture.homeip.net>   Stuart, Ed wrote:n > L > What about if the disk where the OpenVMS security audit file resides fills > up?f >   E You can't log in. So if you don't have a process already logged in totG free up some space, then forcing a crash then doing a minimum boot willo+ let you in to do the necessary cleaning up.o  G Decent monitoring tools should normally provide you with enough warninga0 to take remedial action before you get that far.   -- -
 Paul Sture   ------------------------------  % Date: Fri, 13 Feb 2004 16:48:49 +0100s* From: Paul Sture <nospam@sture.homeip.net>: Subject: Re: [OT] Solaris crashes itself when /tmp is full0 Message-ID: <402CFFF1.465B7F7E@sture.homeip.net>  
 nospam wrote:e > > > in article 402CEBAB.2AD40B83@sture.homeip.net, Paul Sture at4 > nospam@sture.homeip.net wrote on 14/02/2004 01:22: >  > > Stuart, Ed wrote:  > >>O > >> What about if the disk where the OpenVMS security audit file resides fillso > >> up? > >> > > I > > You can't log in. So if you don't have a process already logged in toiK > > free up some space, then forcing a crash then doing a minimum boot will-/ > > let you in to do the necessary cleaning up.  > >oK > > Decent monitoring tools should normally provide you with enough warninga4 > > to take remedial action before you get that far. > N > WRONG answer. When audit is in default or C2 mode the correct response is toL > crash the system if an audit can't be logged. There is no login in to freeK > space its a halt,  not even a friendly shutdown. The author was trying too > point this out.- >   1 Are you sure that's really the default behaviour?3  * From HELP SET AUDIT/SERVER on Alpha V7.3-1  F        FINAL_             Specifies the action the audit server shouldD        ACTION=action      take when it runs out of memory and cannotG                           buffer messages. (For more information, referiF                           to the discussion of message flow control inH                           the OpenVMS Guide to System Security.) Specify7                           one of the following actions:s  B                              CRASH - Crash the system if the audit7                              server runs out of memory.   4C                              IGNORE_NEW - Ignore new event messages0A                              until memory is available. New eventaD                              messages are lost but event messages in.                              memory are saved.  C                              PURGE_OLD (default) - Remove old event:0                              ^^^^^^^^^^^^^^^^^^^G                              messages until memory is available for thee3                              most current messages.e  m     -- m
 Paul Sture   ------------------------------  % Date: Fri, 13 Feb 2004 17:26:14 +0100a* From: Paul Sture <nospam@sture.homeip.net>: Subject: Re: [OT] Solaris crashes itself when /tmp is full0 Message-ID: <402D08B6.63D12D2C@sture.homeip.net>  
 nospam wrote:e > > > in article 402CEBAB.2AD40B83@sture.homeip.net, Paul Sture at4 > nospam@sture.homeip.net wrote on 14/02/2004 01:22: >  > > Stuart, Ed wrote:B > >>O > >> What about if the disk where the OpenVMS security audit file resides fills2 > >> up? > >> > > I > > You can't log in. So if you don't have a process already logged in to K > > free up some space, then forcing a crash then doing a minimum boot will / > > let you in to do the necessary cleaning up.e > >cK > > Decent monitoring tools should normally provide you with enough warninge4 > > to take remedial action before you get that far. > N > WRONG answer. When audit is in default or C2 mode the correct response is toL > crash the system if an audit can't be logged. There is no login in to freeK > space its a halt,  not even a friendly shutdown. The author was trying toI > point this out.  > 	 >  Mark;)u  F Are you quite sure about the default behaviour there? According to theE OpenVMS Guide to System Security, a crash only occurs if you have settF the server using the following command *and* the audit server runs out
 of memory.  % $ SET AUDIT/SERVER=FINAL_ACTION=CRASH   G the documentation quite clearly states that the default for this actionl
 is PURGE_OLD.c  , I am looking at the V7.3-2 documentation at   C http://h71000.www7.hp.com/doc/732FINAL/aa-q2hlg-te/aa-q2hlg-te.HTMld  H The documentation also describes which processes are exempted from beingH suspended when audit log events cannot be written. IOW VMS tries to keepG running while it has the memory resources to buffer unwritten audit logw$ entries, even if FINAL_ACTION=CRASH.     -- t
 Paul Sture   ------------------------------  % Date: Fri, 13 Feb 2004 16:10:06 +0000iO From: Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com>0: Subject: Re: [OT] Solaris crashes itself when /tmp is full/ Message-ID: <c0isv9$5h$1@new-usenet.uk.sun.com>h   Bob Koehler wrote: > In article <c0i7eh$lua$1@new-usenet.uk.sun.com>, Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> writes: > = >>File up the partition that /tmp resides in for example / oni< >>a system without this and you will get the same behaviour. >  > J >    So you are sure Solaris will crash if /tmp is an ordinary file system >    and you fill it up?  Sad.   That wasn't what I said, sad.u  ? I said that on all UNIX systems it would be possible to fill upd1 /tmp, yoou could of course stop this with quotas.   = Whether that causes your UNIX system to crash is debatable, In< ran my server with a full /tmp filesystem last night without8 it crashing (Its a web server) so it was active and then? I cleaned it up in the morning. This was with a memory residento /tmp.i   >  > C >>>b) I've never seen VMS crash because any disk (even system disk)o; >>>filled up (and that's in 22 years managing VMS systems).p >>>t >> >>But other people have. >  > 	 >    Who?  > # Read the rest of the thread thread.E   > ? >>>c) this Solaris, ahem, "feature" has got to be the stupidestvA >>>design I've seen, possibly *ever* in the 35 years I've been inf8 >>>this industry (and I've seen plenty of stupid stuff). >>>e >>= >>Sorry but thats rubbish and if you really don't want to getv= >>the performance hike for using tmpfs then you are perfectly ' >>free to mount /tmp wherever you like.R >  > C >    Any design that allows an ordinary unpriviledged user to crash F >    the system is a design with a security hole.  Trivial DOS attack. > ; So what you have is a problem that might of might not crash48 a Solaris box with a similar problem that might or might not crash an OpenVMS box.l  : It was nice to air these issues. Its also nice to see that0 its the usual OpenVMS glass house stone problem.   regardsa Andrew Harrison.   ------------------------------  % Date: Fri, 13 Feb 2004 16:10:49 +0000 O From: Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com>A: Subject: Re: [OT] Solaris crashes itself when /tmp is full/ Message-ID: <c0it0k$5h$2@new-usenet.uk.sun.com>w   Bill Gunshannon wrote:5 > In article <UiGusAt+zJVs@eisner.encompasserve.org>,o@ > 	koehler@eisner.nospam.encompasserve.org (Bob Koehler) writes: >  >>In article <c0i7eh$lua$1@new-usenet.uk.sun.com>, Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> writes: >>> >>>File up the partition that /tmp resides in for example / on= >>>a system without this and you will get the same behaviour.o >>J >>   So you are sure Solaris will crash if /tmp is an ordinary file system >>   and you fill it up?  Sad. >  > % > I doubt even Solaris would do that.u >   > I didn't say that it would crash, I said you could fill it up.   Regards  Andrew HarrisonS > bill >    ------------------------------  # Date: Fri, 13 Feb 2004 16:48:55 GMTg" From: GreyCloud <mist@Cumulus.com>: Subject: Re: [OT] Solaris crashes itself when /tmp is full7 Message-ID: <Xd7Xb.768$ss.22683@bcandid.telisphere.com>e  ( Andrew Harrison SUNUK Consultancy wrote:   > Bill Gunshannon wrote: > 3 >> In article <402BA1B8.18896617@blueyonder.co.uk>,h= >>     Tim Llewellyn <tim.llewellyn@blueyonder.co.uk> writes:t >> >>>e >>> Roy Omond wrote: >>>e >>>  >>>> Any comments ?I >>>> >>>> A flabbergasted Roy Omond >>>> Blue Bubble Ltd.o >>>e >>>eC >>> Sure, it is a unix "weakness". Some syadmins have cron jobs to s >>> monitor and clean up	 >>> /tmp.o >> >> >> >> Here we go again. >>I >> No, it is not a Unix weakness.  None of the numerous servers I run nowEJ >> will crash if /tmp fills up and none that I remember in the past (whichI >> goes back a couple of decades).  The worst that is likely to happen is-F >> that some application (vi comes immediately to mind) might not work >> properly. >>F >> I am curious enough about this that I plan to set up a machine withG >> Solaris 9 and test this myself.  Even if it turns out to be true, it.G >> is easily remedied and probably would have been, before the fact, by0( >> any really experienced Unix Sysadmin. >> > C > Sun supplies SunMC with Solaris, one of the agents is a diskspacetD > monitor, setting a threshold on /tmp and then finding and deleting6 > all files over a certain age is simple to impliment. > B > As I said earlier this seems to be a people and process problem. >   I Odd.  Seeing that this is an obvious problem, why doesn't the o/s itself iI warn that the limits are being approached?  I noticed that when / itself mC gets full, the o/s emits a FS full message and won't let any users t4 login.  The o/s still stays up under this condition.H Can I set high water marks that the o/s can then emit a warning message?   ------------------------------    Date: 13 Feb 2004 12:58:14 -0600; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler)1: Subject: Re: [OT] Solaris crashes itself when /tmp is full3 Message-ID: <4iIA9pnUGST+@eisner.encompasserve.org>n   In article <c0isit$b$1@new-usenet.uk.sun.com>, Andrew Harrison SUNUK Consultancy <Andrew_No.Harrison_No@nospamn.sun.com> writes: >  > # swap -l  > No swap devices configured >   C    Right.  And when your system pages it probably just goes to that0"    extra memory you weren't using.   ------------------------------   End of INFO-VAX 2004.087 ************************