Oninit cdi / Archive
Talk to an expert →

IDS 7.23 uc1 crash during Update Statistics

Posted in 1999

Topics: Error Codes & Troubleshooting, Versions, Editions & End-of-Life

ssoman@omm.com — — source: Informix-list mailing list archive (1991-1998)

Hi everybody, Our IDS 7.23UC1 crashed this weekend during a routine Update Statistics high of all our tables. It actually crashed on another query that was running during that time. What I noticed was that the offending query was accessing the same table on which an US high was being run. I have a couple of questions on this. 1. Do we need to have all users off the system when we perform an Update Stats ? - We usually do, but of late since the whole process was taking longer than normal, it ran into a few user processes. 2. What is it that happens when an update stats is trying to build the distribution of a table and another query is trying to access it, and has anyone seen a crash like this before ? We do sometimes get errors of "Invalid distribution..." if a query clashes with an update stats. Here is the error msg 19:08:50 Assert Failed: Internal Error - Segmentation Violation 19:08:50 Who: Session(337, mdr07672@la1rs6000, 59122, 1342566996) Thread(531, sqlexec, 5003ec94, 4) 19:08:50 Results: OnLine must abort 19:08:50 Action: Reinitialize shared memory 19:08:50 See Also: /infxdump/af.2139631, shmem.2139631.0 19:10:03 rsinit.c, line 9947, thread 531, proc id 20322, SegmentationViolation . 19:10:03 PANIC: Attempting to bring system down Thanks for your inputs as always. Sujata Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't.

↪ replying to ssoman@omm.com
Art S. Kagel — — source: Informix-list mailing list archive (1991-1998)

That crash should not happen. Probably an old 7.23 bug. I suggest upgrading to 7.31UC3 or later. Art S. Kagel ssoman@omm.com wrote: > > Hi everybody, > Our IDS 7.23UC1 crashed this weekend during a routine Update > Statistics high of all our tables. It actually crashed on another query > that was running during that time. What I noticed was that the > offending query was accessing the same table on which an US high was > being run. I have a couple of questions on this. > 1. Do we need to have all users off the system when we perform an > Update Stats ? - We usually do, but of late since the whole process was > taking longer than normal, it ran into a few user processes. > 2. What is it that happens when an update stats is trying to build > the distribution of a table and another query is trying to access it, > and has anyone seen a crash like this before ? We do sometimes get > errors of "Invalid distribution..." if a query clashes with an update > stats. > > Here is the error msg > 19:08:50 Assert Failed: Internal Error - Segmentation Violation > 19:08:50 Who: Session(337, mdr07672@la1rs6000, 59122, 1342566996) > Thread(531, sqlexec, 5003ec94, 4) > 19:08:50 Results: OnLine must abort > 19:08:50 Action: Reinitialize shared memory > 19:08:50 See Also: /infxdump/af.2139631, shmem.2139631.0 > 19:10:03 rsinit.c, line 9947, thread 531, proc id 20322, Segmentation> Violation . > 19:10:03 PANIC: Attempting to bring system down> > Thanks for your inputs as always. > > Sujata > > Sent via Deja.com http://www.deja.com/ > Share what you know. Learn what you don't.

Glossary