[ZODB-Dev] Conflict Error
Hello everyone, I have a couple questions related to conflict errors. During our home page stress testing (20 concurrent users), we detected several conflict errors (see trace below), We thought these could ONLY appear when writing objects in the ZODB. 2008-05-27T18:57:53 INFO ZODB conflict error at /portal/ (26 conflicts since startup at 2008-05-27T13:37:19) I would like to know: - Can conflict errors appear when reading objects from the ZODB? - Could conflict errors be the cause of a spinning ZOPE scenario? - Is there any situation where ZOPE could try to write objects when rendering an only read page? Our configuration is as follows: Apache + Pound + ZEO architecture (5 ZEO clients, 1 ZEO storage server - 650.000 ZODB objects) Thanks in advance. ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org http://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Conflict Error
Hello Marco, I have this problems too in a not estress testing. In a simple access (2 concurrents users) I have this response: 2008-05-27T13:23:48 INFO ZODB conflict error at /portal/ (22 conflicts since startup at 2008-05-27T10:37:28) 2008-05-27T13:23:48 INFO ZODB conflict error at /portal/ (23 conflicts since startup at 2008-05-27T10:37:28) 2008-05-27T13:29:21 INFO ZODB conflict error at /portal/login_form_usuario_pwd/ (29 conflicts since startup at 2008-05-27T10:37:28) 2008-05-27T13:29:22 INFO ZODB conflict error at /portal/login_form_usuario_pwd/ (30 conflicts since startup at 2008-05-27T10:37:28) 2008-05-27T19:48:40 INFO ZODB conflict error at /portal/portal_css/Plone Default/ploneStyles5658.css (68 conflicts since startup at 2008-05-27T13:37:19) How can they be solved? Can I realize partial commits? How to? Thanks in advance. David 2008/6/5 Marco Bizzarri [EMAIL PROTECTED]: Even though I'm not a ZODB guru, I would suggest investigating if your page is really a read-only one. I had a case with ab tool which I used as a stress tool. It openend a new session for each client and each request, thus potentially causing conflicts. Regards Marco On Thu, Jun 5, 2008 at 12:10 PM, David Otero Figueroa [EMAIL PROTECTED] wrote: Hello everyone, I have a couple questions related to conflict errors. During our home page stress testing (20 concurrent users), we detected several conflict errors (see trace below), We thought these could ONLY appear when writing objects in the ZODB. 2008-05-27T18:57:53 INFO ZODB conflict error at /portal/ (26 conflicts since startup at 2008-05-27T13:37:19) I would like to know: - Can conflict errors appear when reading objects from the ZODB? - Could conflict errors be the cause of a spinning ZOPE scenario? - Is there any situation where ZOPE could try to write objects when rendering an only read page? Our configuration is as follows: Apache + Pound + ZEO architecture (5 ZEO clients, 1 ZEO storage server - 650.000 ZODB objects) Thanks in advance. ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org http://mail.zope.org/mailman/listinfo/zodb-dev -- Marco Bizzarri http://iliveinpisa.blogspot.com/ ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org http://mail.zope.org/mailman/listinfo/zodb-dev
[ZODB-Dev] Conflict Error that won't go away?
Hi there, I'm facing an issue with a ConflictError that won't go away even after restarting. Looks like the ZODB got screwed somehow. Here's the traceback: * Module Zope.App.startup, line 163, in zpublisher_exception_hook * Module ZPublisher.Publish, line 107, in publish * Module Zope.App.startup, line 222, in commit * Module ZODB.Transaction, line 252, in commit * Module ZODB.Connection, line 738, in tpc_vote * Module ZEO.ClientStorage, line 849, in tpc_vote * Module ZEO.ClientStorage, line 833, in _check_serials ConflictError: database conflict error (oid 0xd40acd, serial this txn started with 0x0363a06b42bc2499 2006-02-24 15:39:15.641000, serial currently committed 0x0363b6ce866bdc88 2006-02-28 15:10:31.505000) The system where this is happening is running a ZEO Server and a single ZEO Client. At some point, near the date mentioned in the traceback, the system ran out of disk space during a large transaction. Could this have affected the database? What are the steps to recover from it? Maybe truncate the database? What can be done to avoid it in the future? -- Sidnei da Silva Enfold Systems, LLC. http://enfoldsystems.com ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org http://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Conflict Error that won't go away?
We've seen this problem occasionally. Usually deleting the ZEO cache clears the problem. On Thu, 2 Mar 2006, Sidnei da Silva wrote: Hi there, I'm facing an issue with a ConflictError that won't go away even after restarting. Looks like the ZODB got screwed somehow. Here's the traceback: * Module Zope.App.startup, line 163, in zpublisher_exception_hook * Module ZPublisher.Publish, line 107, in publish * Module Zope.App.startup, line 222, in commit * Module ZODB.Transaction, line 252, in commit * Module ZODB.Connection, line 738, in tpc_vote * Module ZEO.ClientStorage, line 849, in tpc_vote * Module ZEO.ClientStorage, line 833, in _check_serials ConflictError: database conflict error (oid 0xd40acd, serial this txn started with 0x0363a06b42bc2499 2006-02-24 15:39:15.641000, serial currently committed 0x0363b6ce866bdc88 2006-02-28 15:10:31.505000) The system where this is happening is running a ZEO Server and a single ZEO Client. At some point, near the date mentioned in the traceback, the system ran out of disk space during a large transaction. Could this have affected the database? What are the steps to recover from it? Maybe truncate the database? What can be done to avoid it in the future? -- ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org http://mail.zope.org/mailman/listinfo/zodb-dev
Re: [ZODB-Dev] Conflict Error
Antonio Beamud Montero wrote: Where Status is a Persistent Class, with an OOBTree attribute called _dict, and serveral methods wrapping this _dict, like add, remove, etc.. OOBTree's have conflict resolution code which will try and do the right thing instead of raising a ConflictError... When several threads try to add to a the 'completed' dict, a conflict error arises, Well, yes, what did you expect? but if I use distinct keys... Then what? I'd guess the OOBTree conflict resolution code helps you out here :-) why happens? Can be by using the Status class around dict? Sorry, can't understand any of that... A persistent object (inerith from Persistent) has a simple way to force a sync()? No, look at the transaction package for what you want to be doing here... cheers, Chris -- Simplistix - Content Management, Zope Python Consulting - http://www.simplistix.co.uk ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org http://mail.zope.org/mailman/listinfo/zodb-dev
[ZODB-Dev] Conflict Error
Hi all: I have a problem with conflicts in my system. I have implemented a multi-thread server and I have several server as ZEO clients. In ZODB I have created 3 objects: root['pendings'] = Status() root['error'] = Status() root['completed'] = Status() Where Status is a Persistent Class, with an OOBTree attribute called _dict, and serveral methods wrapping this _dict, like add, remove, etc.. When several threads try to add to a the 'completed' dict, a conflict error arises, but if I use distinct keys... why happens? Can be by using the Status class around dict? A persistent object (inerith from Persistent) has a simple way to force a sync()? Thanks. ___ For more information about ZODB, see the ZODB Wiki: http://www.zope.org/Wikis/ZODB/ ZODB-Dev mailing list - ZODB-Dev@zope.org http://mail.zope.org/mailman/listinfo/zodb-dev
