Couldn't load state for 0x013872b7 / AssertionError

[ChrisW]

Same circumstances as #1970, but different error:

10:05 - Couldn't load state for 0x013872b7:
Traceback (most recent call last):
  File "lib/python/ZODB/Connection.py", line
704, in setstate
    self._setstate(obj)
  File "lib/python/ZODB/Connection.py", line
757, in _setstate
    self._load_before_or_conflict(obj)
  File "lib/python/ZODB/Connection.py", line
765, in _load_before_or_conflict
    if not (self._mvcc and self._setstate_noncurrent(obj)):
  File "lib/python/ZODB/Connection.py", line
788, in _setstate_noncurrent
    assert end is not None
AssertionError

[Tim Peters (tim-one)]

More info from Dieter Maurer, at

http://mail.zope.org/pipermail/zodb-dev/2005-December/009623.html

"""
This means that the latest modification for this object
lies before the respective transaction.

In this case, we should not have an invalidation for the object,
such that we would not call "_setstate_current".

I expect a missing "flush_invalidations" during "Connection._setDB".
I had to add such a call in our ZODB version:

    def _setDB(self, odb, mvcc=None, txn_mgr=DEPRECATED_ARGUMENT,
               transaction_manager=None, synch=None):
        ....
        self.transaction_manager = transaction_manager or transaction.manager
        # DM 2005-08-22: always call '_flush_invalidations' as it does
        # more than cache handling only
        self._flush_invalidations()
        if self._reset_counter != global_reset_counter:
            # New code is in place. Start a new cache.
            self._resetCache()
        # DM 2005-08-22: always call '_flush_invalidations'
## else:
## self._flush_invalidations()

"""

[Florent Guillaume (efge)]

FYI we appear to have the same problem in production at a customer site.
I don't know enough of MVCC to dig into this myself.
We'll try Dieter's patch, and report on the effects, and I'll include the patch in ZODB if it proves harmless.

[Florent Guillaume (efge)]

Didn't seem to fix it...

[Jim Fulton (jim-zope)]

Is this still a problem?

[ChrisW (chris-simplistix)]

The guys still working on this project have said they don't see these anymore.

[Jim Fulton (jim-zope)]

On Wed, Sep 15, 2010 at 2:39 AM, ChrisW <email address hidden> wrote:
> The guys still working on this project have said they don't see these
> anymore.

Great! Much thanks for checking.

JIm

--
Jim Fulton

[Jim Fulton (jim-zope)]

This seems to not be a problem any more.