Re: sql causes server crash
Reinhard Habichtsberg said the sysmaster crash persisted even after replacing the round(sum()) expression with 1, suspecting cross-instance sysmaster access. SaltTan reported years of similar use on 9.40 C1, C5, C6 and C7. scottishpoet and Ben Thompson advised trying 9.40.UC8 and then contacting technical support. Unresolved.
Auto-generated by Claude from the posts below — may be imperfect; read the full thread.
> Ben wrote: > Habichtsberg, Reinhard wrote: > > Ben: > > > > If the query is run against _one_ server no error occures. > >Actually I have a > > cron job for this. My intention is to collect statistical > >data from two or > > more instances in one database on one instance. > > > > The problem raises if I mix two server instances: Query > >sysmaster of the > > first and write the result to the second server. > > In which case it sounds like you may not suffering from the same bug. > Even so, have you tried running a similar query without "round > (sum(cumtime/1000000),0)", perhaps just: > > select "$INFORMIXSERVER", current zeitpunkt, 1, reason .... The sql crashes the server even if I replace "round (sum(cumtime/1000000),0)" with "1". I'm nearly sure the crash is associated with the access from the server instance where the data should be inserted to the sysmaster (in other words: shared memory) on the other instance. Reinhard.
Habichtsberg, Reinhard wrote: > > The sql crashes the server even if I replace "round > (sum(cumtime/1000000),0)" with "1". I'm nearly sure the crash is associated > with the access from the server instance where the data should be inserted > to the sysmaster (in other words: shared memory) on the other instance. > > Reinhard. I do select from sysmaster and insert into my database for 3 years every night. I never used 9.40xC4, but it works fine for C1, C5, C6 and C7.
I'd duggest you obtain 9.40.UC8 and see if the problem reproduces there!
Habichtsberg, Reinhard wrote: > The sql crashes the server even if I replace "round > (sum(cumtime/1000000),0)" with "1". I'm nearly sure the crash is associated > with the access from the server instance where the data should be inserted > to the sysmaster (in other words: shared memory) on the other instance. I don't know so I'd just be guessing now. I think your best options are to call technical support or upgrade to 9.40.FC8 and if the problem persists call technical support. Ben.
i suspect technical support will request you see if the problem reproduces in the current fix pack (9.40.UC8)