[ZODB-Dev] BTree corrupted after conflict resolution

[John Belmonte]

Tim Peters wrote:
> [John Belmonte]
> 
>>...
>>My first guess is that it's once again an issue about the server
>>context not being able to locate my key class, and for some reason
>>the exception is getting suppressed.
> 
> I can't guess the details of what you're doing well enough to be of real
> help, so I'll just ask whether you've looked in your ZEO server log file.
> That's where exceptions would get reported, since that's where *attempts* to
> resolve conflicts are made.  If an exception occurs during a
> _p_resolveConflict() method, the actual problem gets logged like so:
> 
>         zLOG.LOG("Conflict Resolution", zLOG.ERROR,
>                  "Unexpected error", error=sys.exc_info())

Yes, I'm certainly looking at the ZEO log, and there is nothing unusual.

> but the exception is not propagated.  ConflictError gets raised instead, no
> matter what the original exception.

I mean *no* error is getting raised, which is why I end up with a 
corrupted BTree in the database.  So whatever operator error this turns 
out to be, it seems there is still a problem with ZODB not raising an 
exception.

-John


-- 
http:// if  ile.org/