[Xapian-discuss] xapian-compact ok, xapian-check failure

[Olly Betts]

On Wed, Jul 20, 2011 at 12:53:59PM +0200, Henry C. wrote:
> I can only assume one of the sub-indexes was bad during the batch run -- and
> bad in such a way that it didn't trip xapian-compact.  Not really
> xapian-compact's fault in my opinion, more an issue of bull-twang concurrency
> checking on my part.

Ah, was one of the sources being updated during compaction?  If so,
that's probably the problem.

We don't really want to just lock the source databases as that would
require them to be writable by the user running the compaction, but we
probably should check we open each table of a particular source at the
same revision.  And an option to lock them might be handy.

Cheers,
    Olly