Receiving "Changing data structure forced command termination." from onstat!

Getting a "Changing data structure forced command termination." message when executing onstat (-g ses, -g ath, and others). oninit is running at ~98% of CPU (via "top"), so system is busy, but can things change that fast?!?! What makes me more nervous is that I receive a "Memory dump (core)" when executing some onchecks (-cc, -ce, -pt), although the -cr worked fine. Any ideas out there, folks?

Just a thought b'coz one time it happened with one of my another DBA. Run "fuser $INFORMIXDIR/bin/*" and check which process thread is running on informix binaries. Check for ontape or onbar binaries. In my case it was a old ontape process which was hanging. When I killed that process, onstat was working fine. Well, it was just a thought. I do not know that why it happen. GOODLUCK!!! Bryan Osborne wrote: > Getting a "Changing data structure forced command termination." message when > executing onstat (-g ses, -g ath, and others). oninit is running at ~98% of > CPU (via "top"), so system is busy, but can things change that fast?!?! > > What makes me more nervous is that I receive a "Memory dump (core)" when > executing some onchecks (-cc, -ce, -pt), although the -cr worked fine. > > Any ideas out there, folks?

Thanks Sanjeev. Unfortunately, fuser didn't show anything. A note of clarification, onstat (-g ses, -g ath, and others) will show a number of threads prior to giving error message: "Changing data structure forced command termination." Anyone other ideas? Anyone else seen this before? TIA... Sanjeev Sagar wrote in message <3869238B.9ED939C2@hotmail.com>... >Just a thought b'coz one time it happened with one of my another DBA. > >Run "fuser $INFORMIXDIR/bin/*" and check which process thread is running on >informix binaries. Check for ontape or onbar binaries. In my case it was a old >ontape process which was hanging. When I killed that process, onstat was working >fine. > >Well, it was just a thought. I do not know that why it happen. > >GOODLUCK!!! > >Bryan Osborne wrote: > >> Getting a "Changing data structure forced command termination." message when >> executing onstat (-g ses, -g ath, and others). oninit is running at ~98% of >> CPU (via "top"), so system is busy, but can things change that fast?!?! >> >> What makes me more nervous is that I receive a "Memory dump (core)" when >> executing some onchecks (-cc, -ce, -pt), although the -cr worked fine. >> >> Any ideas out there, folks? >

This is actually the error handler for a SEGV for onstat. We don't 'freeze' shared memory when run an onstat, but most onstats read from shared memory. Therefor it is possible that the underlying datastructures can be changing while onstat is actually processing the request. For instance, if while running through the thread or session lists the list changes, then onstat will often encountere a SEGV. Instead of aborting, the message "Changing data structure..." is printed. Unless it is consistantly occuring at the same spot, I wouldn't worry too much about it. If it is consistantly occuring at the same spot, then I'd contact tech support. Bryan Osborne wrote: > Thanks Sanjeev. > > Unfortunately, fuser didn't show anything. > > A note of clarification, onstat (-g ses, -g ath, and others) will show a > number of threads prior to giving error message: "Changing data structure > forced command termination." > > Anyone other ideas? Anyone else seen this before? > > TIA... > > Sanjeev Sagar wrote in message <3869238B.9ED939C2@hotmail.com>... > >Just a thought b'coz one time it happened with one of my another DBA. > > > >Run "fuser $INFORMIXDIR/bin/*" and check which process thread is running on > >informix binaries. Check for ontape or onbar binaries. In my case it was a > old > >ontape process which was hanging. When I killed that process, onstat was > working > >fine. > > > >Well, it was just a thought. I do not know that why it happen. > > > >GOODLUCK!!! > > > >Bryan Osborne wrote: > > > >> Getting a "Changing data structure forced command termination." message > when > >> executing onstat (-g ses, -g ath, and others). oninit is running at ~98% > of > >> CPU (via "top"), so system is busy, but can things change that fast?!?! > >> > >> What makes me more nervous is that I receive a "Memory dump (core)" when > >> executing some onchecks (-cc, -ce, -pt), although the -cr worked fine. > >> > >> Any ideas out there, folks? > > -- Madison Pruet =========================================== Enterprise Replication Product Developement Dallas, Texas Informix Software ===========================================

I've seen this occur under HP-UX when shared memory kernel parameters weren't set correctly (see the Release notes directory under INFORMIXDIR for recommendations). I vaguely recall that if "Total Resident memory" allocated to the instance exceeded the shmmax setting at the OS level, the OS needed to create two segments for resident memory at instance initialization. Under this condition, similar "Changing ..." warning messages occurred, especially when the instance was busy. You could check a couple of items - (1) Your online log for warnings about shmmax settings (2) Your actual shmmax settings and compare it to the total Resident size (onstat -g seg). Also, if possible, try experimenting on a test machine. Rudy (Cheers, Scott). Bryan Osborne wrote: > >> Getting a "Changing data structure forced command termination." message > when > >> executing onstat (-g ses, -g ath, and others). oninit is running at ~98% > of > >> CPU (via "top"), so system is busy, but can things change that fast?!?! > >> > >> What makes me more nervous is that I receive a "Memory dump (core)" when > >> executing some onchecks (-cc, -ce, -pt), although the -cr worked fine. > >> > >> Any ideas out there, folks? > >