This is similair to a problem i am experciencing using the DBPool and
PostGresQL. Using the pyPgSQL module, i get connections that begin to go
IDLE in SELECT and IDLE in TRANSACTION when viewed via the 'ps' util.
The DBPool seems to be able to recover from this during the course of
normal operations, however during the weekly full VACUUM everyone seems
to reach a dead lock, effectively hanging the AppServer until it is kill
9'ed and restarted.
Any ideas on how to work around this?
Aaron Held wrote:
> I had the same problem with pgSQL a little over a year ago. I'm a fan
> of Postgres and have had million row tables, but you will get the same
> connection issue.
> Frank Barknecht wrote:
>> Thomas E Jenkins hat gesagt: // Thomas E Jenkins wrote:
>>> On Mon, 2003-11-10 at 05:25, Frank Barknecht wrote:
>>>> Tried now, but it didn't help. I might have to live with this problem,
>>>> but then I would need a way to restart the AppServer with a "kill -9
>>>> allAppServerPIDs" beforehand automatically...
>>> The only way I found to prevent these stale connections is to
>>> build a new one for every group of DB operations. That of course
>>> away any connection pooling you can do. You may also want to look at
>>> David Rushby's resilient DBPool and see if you can make more progress
>>> than I in implementing a MySQL version. The message can be found here:
>> Well, I'm using SQLObject, so I'm dependent on SQLObject's pooling
>> (maybe this can be switched off, though, I didn't look). I'm really
>> considering to switch to Postgresql now, which is possible thanks to
>> the abstraction provided by SQLObject, but it still would be a major
>> step. I'm wondering, if this could help, or if I'd just take my
>> problems to another backend... ?!
> This SF.Net email sponsored by: ApacheCon 2003,
> 16-19 November in Las Vegas. Learn firsthand the latest
> developments in Apache, PHP, Perl, XML, Java, MySQL,
> WebDAV, and more! http://www.apachecon.com/
> Webware-discuss mailing list