[Labs-l] Database corruption

Tue May 24 21:52:20 UTC 2016

Russel,

Thanks for raising this concern.  I don't know what the cause may be, but I
will try to loop in our database administrators.  If you could add some
description and details on the more recent corruption suspected in T134203
that would be helpful.

Chase Pettet

On Tue, May 24, 2016 at 4:05 PM, Russell Blau <russblau at imapmail.org> wrote:

> Is database corruption considered a priority issue for Labs? The enwiki
> database replica has been corrupt since about 30 April; I posted a bug
> report[1] on 2 May, but it hasn't even been triaged yet. In the meantime,
> there seems to have been further corruption within the past 24-36 hours,
> as evidenced by tools that report link counts (based on the replica) that
> are highly inaccurate when checked on the live wiki.
>
> I'm surprised this isn't causing problems for other users' tools (or at
> least that no one has complained about it). At a minimum, I know that the
> pagelinks table in the replica contains entries that don't exist on the
> live wiki. I would assume it likely is also missing some entries that do
> exist on the live wiki, although this is much harder to confirm. Other
> tables may be inaccurate, as well.
>
> [1] https://phabricator.wikimedia.org/T134203
>
>
> _______________________________________________
> Labs-l mailing list
> Labs-l at lists.wikimedia.org
> https://lists.wikimedia.org/mailman/listinfo/labs-l
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.wikimedia.org/pipermail/labs-l/attachments/20160524/82e1ea0f/attachment.html>