[Geoserver-users] PostgreSQL connections

what's the status of the "too many postgresql connections" problem? I'm
using 1.5RC3 and still seeing this issue; the lastest stuff on this list is
from middle of last year.

thanks,
mike
--
View this message in context: http://www.nabble.com/PostgreSQL-connections-tf3567655.html#a9966419
Sent from the GeoServer - User mailing list archive at Nabble.com.

Hi Mike,

From what I know this should not be a problem in 1.5.x series. Can you give us a bit more information about what types of requests you are making?

Thanks,

-Justin

mfrumin wrote:

what's the status of the "too many postgresql connections" problem? I'm
using 1.5RC3 and still seeing this issue; the lastest stuff on this list is
from middle of last year.

thanks,
mike

--
Justin Deoliveira
The Open Planning Project
http://topp.openplans.org

mfrumin ha scritto:

what's the status of the "too many postgresql connections" problem? I'm
using 1.5RC3 and still seeing this issue; the lastest stuff on this list is
from middle of last year.

Long story short, we weren't able to reproduce your issues, not even with a simple stress test I used last year, that's why nothing has been
done.

If you can provide us with test data and a set of requests that
repeated over time cause the problem we would be eager to solve it.

Btw, do you see in your log messages such as:
12 avr. 2007 16:11:46 org.geotools.data.jdbc.QueryData finalize
GRAVE: There's code leaving readers, writers or iterators unclosed (you
got an unclosed QueryData object, which is usually held by a reader or a
writer).
Call reader/writer.close() or FeatureCollection.close(iterator) after
using them to ensure they do not hold state such as JDCB connections.
QueryData was open against feature type: parcelles_polygon

I added these messages specifically to track connection lost
due to code not releasing the connection to the pool.

Also, how many concurrent connection are allowed to your server?
I looked in the archives, but all threads about this issue end
without providing enough information to reproduce the problem.

Cheers
Andrea