[Zope] Zope=ZEO connection

[Dennis Allison]

Arggggh...
I forgot that internal logging levels override.
Still, you'd think something that triggers a crash would have a level
higher than info.  Careful examination of the trace log shows nothing.
I'll fix the log level & see if that turns anything up.   


On Tue, 7 Feb 2006, Paul Winkler wrote:

> On Tue, Feb 07, 2006 at 06:18:49PM -0800, Dennis Allison wrote:
> > 
> > Zope 2.9.0
> > Python 2.4.2
> > Amd dual Opteron
> > Linux
> > 
> > I am seeing occasional (several per day)  Zope crashes under load.  The 
> > event.log immediately before and after the crash looks like:
> > 
> > 
> > 2006-02-06T14:07:20 INFO ZPublisher.Conflict ConflictError at 
> > <some path>: 
> > database conflict error (oid 0x086e, class BTrees._OOBTree.OOBTree, serial 
> > this txn started with 0x03633ca95f75e900 2006-02-06 22:01:22.373575, 
> > serial currently committed 0x03633caf59114244 2006-02-06 22:07:20.875176) 
> > (463 conflicts (0 unresolved) since startup at Mon Feb  6 08:21:26 2006)
> > ------
> > 2006-02-06T14:09:09 INFO ZServer HTTP server started at Mon Feb  6 
> > 14:09:09 2006
> >         Hostname: x-harper
> >         Port: 8081
> > 
> > which also triggers a connection drop and restarBt of ZEO.
> > 
> > The failure happens silently even with 
> > 
> > <eventlog>
> >   level ALL
> >   <logfile>
> >     path $INSTANCE/log/event.log
> >     level info
>       ^^^^^^^^^^
> 
> fwiw, you're still only logging at INFO.
> 
> > What sort of Zope failure can cause this sort of behavior?  What's the 
> > best approach to get more information to localize the failure.
> 
> I really don't know. But I'd have a look at the trace log,
> if you've got that configured (I routinely do).
> Maybe you could spot some patterns right before the crashes?
> *shrug*
> 
> 

--