[Zope] RE: [ZODB-Dev] can't pack ZODB anymore

Tim Peters tim at zope.com
Tue Jul 27 10:28:07 EDT 2004


[shan sverige]
> When I try to pack I get:
>
>    Error Value: /usr/local/Zope-2.7.0/var/Data.fs:31906881:time-stamp
>    reduction: 03537a82059cf1cc <= 0355bcc95b5ace33

Ouch.  Looks like the system clock on your machine got set backwards by
about 3 months at some point:

>>> from ZODB.utils import *
>>> print TimeStamp.TimeStamp(p64(0x03537a82059cf1cc))
2004-03-02 17:38:01.315562
>>> print TimeStamp.TimeStamp(p64(0x0355bcc95b5ace33))
2004-06-12 12:57:21.411260
>>>

Then you got bit by this:

    http://zope.org/Collectors/Zope/1327

While no distinct ZODB release was made to correspond with Zope 2.7.2, that
bug should be fixed in Zope 2.7.2.  It's still a panic-level problem if the
clock is seen to go backwards by that much, but at least a panic message
will get logged about it now, and FileStorage will force transaction ids to
be monotonically increasing anyway now.

In any case, packing isn't an error recovery procedure.  Run the diagnostic
scripts explained here, and start making backups (also explained here):

    http://zope.org/Wikis/ZODB/FileStorageBackup

> After running fsrecover.py the resulting Data.fs can't be mounted. I now
> get:
>
>    ZODB.POSException.ConflictError: database conflict error

Sorry, but I bet you have multiple problems in your database now, and
probably stemming from whatever went wrong with your system that caused your
clock to go insane.  This could be difficult to recover from.  Running
fstest and fsrefs (explained on the page linked to above) will help
determine the extent of the damage.



More information about the Zope mailing list