[OpenVMS Alpha V8.3] Crash in SYS$EIDRIVER

Today we had our 4th crash on 4 different MARVELs with the same symptom: It seemingly occured hours/days after a network disconnect (switch poweroff) and showed SYS$EIDRIVER as the culprit CPU bugcheck codes: CPU 001 -- database address 83254300 -- INVEXCEPTN, Exception while above ASTDEL .... Exception Frame at FFFFFFFF.A2945C40 ------------------------------------ R2 = FFFFFFFF.8222B700 SYS$EIDRIVER+0E500 R3 = FFFFFFFF.A2EAC000 R4 = FFFFFFFF.8341044C R5 = 00000000.00000000 R6 = FFFFFFFF.8222BCD8 SYS$EIDRIVER+0EAD8 R7 = FFFFFFFF.A2EAC000 PC = FFFFFFFF.8070C2B8 SYS$EIDRIVER+042B8 PS = 20000000.00001504 MBZ SPAL MBZ IPL VMM MBZ CURMOD INT PRVMOD 0 20 00000000000 15 0 0 KERN 1 KERN Instruction Stream: ------------------- SYS$EIDRIVER+042A8: LDL R0,#X010C(R16) SYS$EIDRIVER+042AC: LDAH R16,#XFFFF(R31) SYS$EIDRIVER+042B0: LDA R16,#X7FFF(R16) SYS$EIDRIVER+042B4: LDL R1,#X0008(R0) PC => SYS$EIDRIVER+042B8: LDL R1,#X0050(R1) SYS$EIDRIVER+042BC: LDL R26,(R1) SYS$EIDRIVER+042C0: SRL R26,#X0F,R27 SYS$EIDRIVER+042C4: BLBC R27,#X000023 SYS$EIDRIVER+042C8: LDQ_U R31,(SP) HPQ provided us a new (X-55) SYS$EIDRIVER from 7-FEB-2008 (!) to replace the one (X-54) from vanilla OpenVMS Alpha V8.3 (11-JAN-2008) to fix it (still needs to be proved ;-) Isn't it sad, that this 21mon old patch is still not a valid VMS83A_LAN ECO? You're now warned (and don't need to multiply crash to want this image ;-) -EPLAN PS: The VMS83A_UPDATE V11 is still not released (was promised for End Sep09) -- Peter "EPLAN" LANGST�GER Network and OpenVMS system specialist E-mail Peter@LANGSTOeGER.at A-1030 VIENNA AUSTRIA I'm not a pessimist, I'm a realist

On a similar but not directly related topic..... I have recently discovered: 1. After the installation of the recent SYS (V8) kit for V8.3-1H1, IBM MQ will hang. A new PROCESS_MANAGEMENT.EXE is available from HP. 2. After the installation of the recent PTHREAD (V1) kit for V8.3-1H1 Oracle CRS would not work. No fix is currently available, uninstall the patch. fwiw, Guy Peleg President Maklee Engineering http://www.maklee.com

On 24 =D0=BD=D0=BE=D1=8F, 14:30, Guy Peleg <makleeengineer...@gmail.com> wr= ote: > > 2. After the installation of the recent PTHREAD (V1) kit for V8.3-1H1 > Oracle CRS would not work. No fix is currently available, > uninstall the patch. > Guy, what were the symptoms that "CRS would not work"? I checked with one of my installations: OpenVMS V8.3-1H1; last ECO+last SYS+PTHREAD-V0100; Oracle 10.2.0.4.2 RAC is working so CRS is OK. Seems that no fix is needed actually. Best regards, Kirill

Hello Guy, > On a similar but not directly related topic..... > > I have recently discovered: > > 1. After the installation of the recent SYS (V8) kit for V8.3-1H1, IBM > MQ will hang. A new PROCESS_MANAGEMENT.EXE > is available from HP. What version of MQ are you running? I ask because a site I manage is contmplating a move to Integrity. Thanks, Jeremy Begg

Jeremy Begg <jeremy@vsm.com.au> writes: >What version of MQ are you running? I ask because a site I manage is >contmplating a move to Integrity. I'm not Guy, but we're running the MQ Client V6.0 on an Integrity rx3600 running V8.3 with no problems. We haven't installed the sys kit with the issue since it's for V8.3-1.

On Nov 26, 3:16=A0am, Jeremy Begg <jer...@vsm.com.au> wrote: > Hello Guy, > > > On a similar but not directly related topic..... > > > I have recently discovered: > > > 1. After the installation of the recent SYS (V8) kit for V8.3-1H1, IBM > > MQ will hang. A new PROCESS_MANAGEMENT.EXE > > is available from HP. > > What version of MQ are you running? =A0I ask because a site I manage is > contmplating a move to Integrity. > > Thanks, > > =A0 =A0 =A0 =A0 Jeremy Begg @Guy: there must be something else different about the system with Oracle CRS that you looked at as other customers have it successfully running with OpenVMS V8.3-1H1; PTHREAD-V0100 is installed as well as last ECO and SYS. Oracle version is 10.2.0.4.2 RAC.

In article <eae427ff-06fc-4e0f-80bd-0b2f0f6afd43@e31g2000vbm.googlegroups.com>, Guy Peleg <makleeengineering@gmail.com> writes: >On a similar but not directly related topic..... > >I have recently discovered: > >1. After the installation of the recent SYS (V8) kit for V8.3-1H1, IBM >MQ will hang. A new PROCESS_MANAGEMENT.EXE >is available from HP. A SYS V9 is also available (though it is not in the master eco list) -- Peter "EPLAN" LANGST�GER Network and OpenVMS system specialist E-mail Peter@LANGSTOeGER.at A-1030 VIENNA AUSTRIA I'm not a pessimist, I'm a realist

On Nov 27, 5:15=A0am, pe...@langstoeger.at (Peter 'EPLAN' LANGSTOeGER) wrote: > In article <eae427ff-06fc-4e0f-80bd-0b2f0f6af...@e31g2000vbm.googlegroups= ..com>, Guy Peleg <makleeengineer...@gmail.com> writes: > > >On a similar but not directly related topic..... > > >I have recently discovered: > > >1. After the installation of the recent SYS (V8) kit for V8.3-1H1, IBM > >MQ will hang. A new PROCESS_MANAGEMENT.EXE > >is available from HP. > > A SYS V9 is also available (though it is not in the master eco list) > > -- > Peter "EPLAN" LANGST=D6GER > Network and OpenVMS system specialist > E-mail =A0Pe...@LANGSTOeGER.at > A-1030 VIENNA =A0AUSTRIA =A0 =A0 =A0 =A0 =A0 =A0 =A0I'm not a pessimist, = I'm a realist MQ involved was V6.0 As for Oracle, the problem I have seen was crashes of the CRSD process. It will not crash every startup, only every now and then. I have no doubts that it's the fault of the PTHREAD kit as multiple times I have removed it and re-installed it and proved the problem shows up only with the kit. Keep in mind that every RAC installation is different, the number of databases, services and listeners is most probably related, although I'm not sure. Another interesting feaure I encountered recently was SHOW PROC/ID=3Dxxx/ CONT may result in a system crash. VMS831H1I_LIBOTS2-V0200 addresses the problem. Guy

On Nov 27, 3:15=A0pm, pe...@langstoeger.at (Peter 'EPLAN' LANGSTOeGER) wrote: > In article <eae427ff-06fc-4e0f-80bd-0b2f0f6af...@e31g2000vbm.googlegroups= ..com>, Guy Peleg <makleeengineer...@gmail.com> writes: > > >On a similar but not directly related topic..... > > >I have recently discovered: > > >1. After the installation of the recent SYS (V8) kit for V8.3-1H1, IBM > >MQ will hang. A new PROCESS_MANAGEMENT.EXE > >is available from HP. > > A SYS V9 is also available (though it is not in the master eco list) > > -- > Peter "EPLAN" LANGST=D6GER > Network and OpenVMS system specialist > E-mail =A0Pe...@LANGSTOeGER.at > A-1030 VIENNA =A0AUSTRIA =A0 =A0 =A0 =A0 =A0 =A0 =A0I'm not a pessimist, = I'm a realist The SYS V9 kit fixes the above mentioned issue of MQ hang.

In article <511f79e2-d308-4e84-a899-305cc878e79c@m3g2000yqf.googlegroups.com>, Volker Halle <volker_halle@hotmail.com> writes: >now there is an VMS83A_UPDATE-V1100.ZIPEXE file available since 30- >NOV-2009, but there is NO related VMS83A_UPDATE-V1100.TXT file on the >FTP site. In Europe it was 1-DEC-2009 (on 30-NOV-2009 only the master list was there) when the kit appeared, and I installed it already on my 1st box (yesterday). But it doesn't contain the VMS83A_LAN V4 (with the SYS$EIDRIVER image) because it still doesn't exist. So, my wait for UPDATE V11 (since Sep09) is over. Good. (funny, UPDATE V12 was expected for Dec09, is now said to appear Jan10) But for the newer SYS$EIDRIVER.EXE you still need to call HPQ or wait... -EPLAN PS: Most of the information in the missing UPDATE V11 ECO Cover Letter is contained in the UPDATE V11 ECO Release Notes contained in the kit. Extract and read them before installing the UPDATE kit (as usual)... -- Peter "EPLAN" LANGST�GER Network and OpenVMS system specialist E-mail Peter@LANGSTOeGER.at A-1030 VIENNA AUSTRIA I'm not a pessimist, I'm a realist

Just for the records VMS83A_UPDATE V12 is out (since 16/17 DECEMBER 2009). It (still) doesn't contain the SYS$EIDRIVER.EXE fix (au contraire, it replaces the fixed image of Feb-2008 with the vanilla V8.3 one of Jan-2008 - forcing the reappearance of the crash) There is still no official VMS83A_LAN (V4 or later) with this image. Sigh -EPLAN PS: next VMS83A_UPDATE ECO is announced for (the week of) 30-APR-2010 -- Peter "EPLAN" LANGST�GER Network and OpenVMS system specialist E-mail Peter@LANGSTOeGER.at A-1030 VIENNA AUSTRIA I'm not a pessimist, I'm a realist