Our setup is: Zope 2.6, Python 2.1.3, ZEO from CVS (http://cvs.zope.org/ZEO/?only_with_tag=ZODB3-3_1-branch)
This morning we encountered a baffling phenomenon. A cron job routinely packs Data.fs by calling the appropriate URL via wget.
52 minutes after packing was completed we noticed a rapid succession of POSKeyError messages in our logs. Attached are 2 typical ones. Note that every traceback from ZODB.FileStorage.modifiedInVersion was directly follow by a (corresponding?) traceback in ZEO.zrpc.connection.call. Also every POSKeyError was transient. Enough patience in reloading the requested page always resulted in a complete rendering.
Stopping Zope and ZEO processes and restarting them "solved" the problem completely.
Any guesses? Perhaps a ZEO-version more appropriate for Zope 2.6?
-------------------------------------- 2002-11-28T08:40:40 INFO(0) zrpc:23837 zeoLoad() raised exception: 0000000000028438 Traceback (innermost last): Module ZEO.zrpc.connection, line 226, in handle_request Module ZEO.StorageServer, line 308, in zeoLoad Module ZODB.FileStorage, line 663, in modifiedInVersion POSKeyError: 0000000000028438 2002-11-28T08:40:40 ERROR(200) ZODB Couldn't load state for '\x00\x00\x00\x00\x0 0\x02\x848' Traceback (innermost last): Module ZODB.Connection, line 507, in setstate Module ZEO.ClientStorage, line 516, in load Module ZEO.ServerStub, line 73, in zeoLoad Module ZEO.zrpc.connection, line 322, in call POSKeyError: 0000000000028438 --------------------------------------
Jo.