From: Olly Betts <olly <at> survex.com>
Subject: Re: Possible disk problems on the search machine
Newsgroups: gmane.discuss
Date: 2008-05-16 16:19:39 GMT (4 years, 16 weeks, 4 days, 12 hours and 41 minutes ago)

On 2008-05-14, Olly Betts <olly <at> survex.com> wrote:
> On 2008-05-13, Olly Betts <olly <at> survex.com> wrote:
>> forever, certainly for days) after reporting several partitions.
>> My guess is it is hanging on the NFS mounts.
>
> Still got these df processes though.

They've all gone now.

>> I can attempt to restart indexing at the point it died, but it will
>> require a bit of coding to enable this to work.  Meanwhile, the live
>> search index won't be getting new articles I'm afraid.
>
> It's indexing the remaining documents now.

OK, everything is rebuilt and running now.  Sorry it took so long.

The live index is currently missing about the last week's worth of
article.  The spooler is busily catching up - depending how long that
takes, I might kick off the next incremental update by hand, or just
leave it to the cron job (early morning Norwegian time).

We're now running on a development snapshot of Xapian using the new
development backend "chert".  This has knocked 44% off the size of
the postlist table (even I'm amazed by how much difference this has
made).  The postlist table is the one which we want to be able to
cache lots of, so this should mean that searches typically run a bit
faster than before.

Cheers,
    Olly