On Fri, Oct 03, 2003 at 12:22:54AM +0200, Dieter Maurer wrote:
Paul Winkler wrote at 2003-10-1 13:28 -0400:
Anybody had problems with zope 2.6.2 not responding for a while? Also running ZEO from ZODB 3.1.2.
it happened this morning for 20 minutes, then spontaneously resolved itself; same thing appears to be happening now.
Do you know what your hanging requests are doing?
not yet. I'm currently going through the big-M log trying to see a pattern.
We had an issue with Postgres where suddenly (harmless) queries took up to an hour to complete. Then several dozens of long running queries completed within a few seconds. The problem disappeared after we upgraded the Postgres server.
not an issue here... there's no database involved other than the ZODB. I have wondered if maybe it's a ZEO issue, or a networking issue between zope & the zeo server. Seems unlikely given that they are in the same rack, conneted via a 100baseT ethernet switch, and ping time between them is very fast every time i've tried it. The problem was noticed after upgrading the server to Zope 2.6.2 / ZEO 2 (from ZODB3-3.1.2.) I downgraded back to Zope 2.5.1 / ZEO 1, which seems to have somewhat reduced the severity of the problem, but it's still there. Now the "stuck" periods are 0 - 3 minutes instead of 20 minutes or more. It may be that this was happening all along and I never noticed before due to the short duration of the outages. I have not seen this on any of our other systems. Sigh. -- Paul Winkler http://www.slinkp.com Look! Up in the sky! It's THE AMAZING HEATHEN I LOVE YOU! (random hero from isometric.spaceninja.com)