Hi!
We use postgresql-6.5.2, on i386-linux (Rh.6.1), from 4-5 Xterminal, the
apllications, and postgres runs on the same machine.
there are approximately 30-40 inserts|updates /min.
The performance is pertty well... but there are many backend crash
(~10/day) ;(
And after a crash some dust remains: uninitialized pages in tables, wrong
indices..., and this dust generate a crash again. (if i not run a vacuum)
I tried, to dump-initdb-restart, but it doesn't help.
We use many user-defined functions (some C, but mostly plpgsql)
this under test circumstances (on another machine) works fine, and i trust
in postgres stability.
So:
Meet somebody with similar things?
How can i find the reasons posteriorly?
Can be the cause of the random backend smash broken hardware?
Other advices?
Thanks...