[ZODB-Dev] Site crash and Data.fs truncated

[Ramon Aseniero]

Hi Tim,

Could this be a bug in Zope? I am currently running zope 2.6.4 (source
release, python 2.1, linux2), python 2.2.3, linux2) on Debian.

Could this be also due to a memory leak on my code?

Thanks,
Ramon

-----Original Message-----
From: Tim Peters [mailto:tim at zope.com] 
Sent: Monday, May 24, 2004 6:38 PM
To: 'Ramon Aseniero'; ZODB-Dev at zope.org
Subject: RE: [ZODB-Dev] Site crash and Data.fs truncated

[Ramon Aseniero]
> My Data.fs was suddenly truncated and it created a Data.fs.tr0 file.

You should look in your log files, for ZODB FS WARNING msgs.  One of
those
is logged whenever a .tr<digit> file is created.  This only happens when
the
FileStorage index is being read, and that happens when the database is
first
opened.  The code reading the index determines that the .fs file is
damaged,
and chops it off at the point of damage.  So the .tr0 file isn't the
damage
itself, it's the result of earlier damage.  The .tr0 file contains
everything that was in the original .fs at and after the damaged point.

> ...
> Module ZODB.FileStorage, line 623, in _load 
> error: unpack str size does not match format

That's peculiar too, since the part *retained* in the .fs file should
contain whole records.

> Any idea?

Sorry, not really.  As you've already discovered, there's not much to go
on.

> How can I fix it?

Revert to a backup at a known good state, if you have one.  If not,
running
fsrecover.py may or may not yield a "good enough" repaired .fs file.