[Gluster-devel] Gerrit down for 10 to 15 minutes for reindexing

[Michael Scherer]

Le lundi 25 janvier 2016 à 15:41 +0100, Michael Scherer a écrit :
> Hi,
> 
> in order to fix some issues (I hope), I am gonna start a reindex of the
> lucense DB of gerrit. This requires the server to be put offline for a
> while, and I did a test on another VM, would take ~10 minutes (it was
> 240 seconds on the VM, but it was likely faster since the VM is faster)
> 
> I will to do that around 18h UTC, in ~ 3h ( so 1pm Boston time, 23h Pune
> time, and 19h Amsterdam time, so it shouldn't impact too much people who
> would either be sleeping and/or eating ). 
> 
> If people really want to work, we always have bugs to triage :)

So it took 3 to 4 minutes, much faster than what I tought.

And it did fixed the issue it was meant to fix, ie Prashanth being
unable to find his own reviews.

This did happen because we did some modification directly in SQL, but
gerrit need to reindex everything to see some changes, and this requires
to take the db offline (something that is fixed with a newer version of
gerrit).

Please ping me if anything weird appear.
-- 
Michael Scherer
Sysadmin, Community Infrastructure and Platform, OSAS


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part
URL: <http://www.gluster.org/pipermail/gluster-devel/attachments/20160125/af91a1d8/attachment.sig>