[Labs-l] Labs testing/bug triage (was "Bot cluster reliability issues")

[Ryan Lane]

On Mon, Feb 4, 2013 at 4:22 AM, Sumana Harihareswara
<sumanah at wikimedia.org>wrote:

>
> It'd be good if there were a clear list of what the most serious issues
> are, so I started looking in Bugzilla.
>
> I see that we have about 106 open issues in Bugzilla against various
> components in Labs[0], including 43 non-enhancement bugs against
> Infrastructure/General/Other[1].  Are those prioritized right now in
> about the right priority, reflecting what Ryan and Andrew want to be
> working on next?  Does the list need bug triage?  Since some of the bug
> reports are several months old, would it be useful for someone to go
> through and check old ones for reproducibility?
>
>
We don't have a documented priority list, currently. Since it's only
Andrew, Mike and myself and our general skillsets cover different things,
we've been handling this on the fly.

Mike has been doing bug wrangling, shell requests, project requests, etc.
Andrew has been handling labsconsole interface bugs and general usability
bugs as of late. I've been working on stabilization and performance
improvements as of the last couple weeks.

I've begun organizing bugs into projects:

<
http://www.mediawiki.org/wiki/Wikimedia_Labs/Account_creation_improvement_project
>
<
http://www.mediawiki.org/wiki/Wikimedia_Labs/Instance_creation_improvement_project
>
<
http://www.mediawiki.org/wiki/Wikimedia_Labs/Interface_usability_improvement_project
>

In general we determine priority of the tasks ourselves and tackle the ones
we think will have the most impact.

Since we'll be adding a number of people into the mix soon, we should have
a weekly meeting and start handling ticket triage in a saner way. I'm a fan
of letting people determine priority themselves. If anyone has a concern
that a specific bug isn't being addressed they should be raising a flag on
the list and we'll discuss re-prioritization there, in-between meetings.

For meetings themselves, I'd like to have open google+ hangouts, with the
ability for all volunteers to join in and raise concerns.

(And btw, when should we be reporting issues in the "General" component
> versus "Infrastructure"?)
>
>
General should be used if people don't know where to put a ticket.
Infrastructure is for reporting bugs against things that affect all
projects, like glusterfs or LDAP issues. A bug for adding replicated DB
support, for instance, would go under Infrastructure.

For bugs specific to bots or tool labs that don't affect other projects we
should have a component for those bugs.

- Ryan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.wikimedia.org/pipermail/labs-l/attachments/20130204/1f672cbe/attachment.html>