Maybe a bit too clever in the Postgresql usage ("In case I want to do what I never want to do" bites me...)
Last update on .
Fighting a bug in a billing process most of the afternoon. Basically two errors, one a memory error and one a database buffer exhaustion error. The first I haven't tracked down, but I did work around it. The second appears to be a side effect of a never-used set of features in PyTable that allow for easy GUI editing of result-sets.
Basically the code was keeping around references to cursors (result-sets) instead of connections, so it appears to have been keeping a large number of buffers active on the server. No problem on the development or testing servers (which have modern postgresql with lots of buffers), but seems to kill the production server. Testing will see if the problem is now solved.
Pingbacks are closed.
Comments are closed.