Re: [sup-talk] DatabaseCorruptError

[Andrew Pimlott <andrew@pimlott.net>]

Excerpts from William Morgan's message of Tue May 10 17:36:54 -0700 2011:
> Reformatted excerpts from Andrew Pimlott's message of 2011-05-06:
> > Sup had been running for a while, and I just came back to my terminal to
> > find it dead:
> > 
> > --- IOError from thread: periodic poll
> > DatabaseError: Error reading block 4279173482: got end of file
> > I'm running an old git pull from next, commit 1bafcf6 dated 2010-10-12.
> 
> Yipes. Are you out of disk space, or anything like that?

I wasn't out of space--unless possibly Xapian caused some transient
spike in disk use that was freed when it died.  The system does run
without a ton of free space though, maybe 100 MB for a Xapian database
that is ~300 MB total.

I didn't get any disk errors either.  This is on a VPS that could
conceivably be hiding some errors.

> > Anything I should do to try to diagnose and recover this?  Or just
> > load from sup-dump?
> 
> Not knowing very much about Xapian, I suggest sup-dump. Not a great
> solution, I know.

It turned out that sup-dump still worked.  Probably some index was
corrupt but the data could still be walked.  So I was able to restore
without losing any data.  (Well, it did use most of the 160 MB of memory
on the VPS.)

> Wait for heliotrope to solve your problems (or at
> least be something I understand and can fix).

Dum de dum dum...

Andrew
_______________________________________________
sup-talk mailing list
sup-talk@rubyforge.org
http://rubyforge.org/mailman/listinfo/sup-talk