Brad Anderson wrote:
Stephen A. Cochran Lists wrote:
Is there a ticket number?
I was not sure if the issue was crappy code in my plugins, which use a central
database across all projects. Unless you're using the DbAuthPlugin, then we
can maybe eliminate that, and know that the problem is in Trac (as well as my
plugins...)
Go ahead and start a ticket, if you want.
I have a system with 14 projects set up, each with it's own database
and username. So within a few days every httpd process will be holding
a db connection for every trac project it's served out. Restarting
apache seems to close all the database connections, so that's an easy fix.
We have 70 projects, but only about 25 are active and using Trac. And indeed,
on an Apache restart, everything is fine, but it takes a while to clean up the
Apache threads and the idle connections.
I wonder if this is because mod_python can't reuse the connection to
different databases. I have been using PostgreSQL and haven't
experienced this problem yet. However, I am using my schema patch so
that all my trac sites are in one database but separate schemas.
http://trac-hacks.org/wiki/PgSchemaPatch
-John
_______________________________________________
Trac mailing list
[email protected]
http://lists.edgewall.com/mailman/listinfo/trac
