Forum OpenACS Development: Deadlocks in postgres

Collapse
1: Deadlocks in postgres
Posted by Malte Sussdorff on 05/10/06 03:54 PM
In one of our sites we see deadlocks on a regular basis, though we are unsure how to find out where they occur (and why). Additionally I was wondering if it would not make sense if postgresql returns a deadlock that the OpenACS database api tries to run the statement that was blocked again like one second later, assuming the deadlock has removed itself by then.
Collapse
2: Re: Deadlocks in postgres (response to 1)
Posted by Andrew Piskorski on 05/11/06 11:39 AM
Dunno, what does PostgreSQL tell you abou the deadlock? Hopefully it dumps some sort of stack trace showing you the SQL the deadlocked transaction was doing?

I'm not familiar with how PostgreSQL handles deadlock, although I was worrying about such things for Oracle back in Feb. 2001, and got some interesting answers. That link seems to broken unfortunately, but it is still on the Google cache.

Collapse
3: Re: Deadlocks in postgres (response to 1)
Posted by Jun Yamog on 05/25/06 01:01 AM
Anyone ever got the contents of the old bboard? It would have been nice if RedHat gave the aD bboard archive to OpenACS.