Re: [HACKERS] Updated 6.5 HISTORY

Oleg Bartunov <oleg(at)sai(dot)msu(dot)su> writes:
> Today I tried my test script (many tables joining) and got a crash.
> It works several days before. CPU usage shows only 5% loading during
> this test. what postgres is doing ? Hmm, just rerun the test and it
> works :-) OK, one more again and it crashed. Again - works. Again -
> crashed, crashed, crashed. Very unstable situation.

Oh dear ... can you provide a backtrace from one of the crashes?

> Does anybody tried sqlbench, posted by
> Edmund ? I tried several times on Linux box, FreeBSD box with current cvs
> and it never finished. RAM usage during the test was about 11-13 Mb
> and CPU usage was extremly low - 5-9% only !

It ran through to completion for me, but took hours. (Unfortunately
I only have access to one machine that has 700Mb of free disk space,
and I mustn't shut off the data collection task that is its primary load.
So I'm stuck with taking a long time to try the sqlbench stuff...
it does seem to work, but I can't say much about performance...)

regards, tom lane