[Labs-l] Slow steps ahead to merge bots and tools

[Petr Bena]

Okay, no doubt, it's easier for you to build the cluster in separate
project - but it will be far harder for bot operators to move all
their tools and databases across project.

I don't really see many problems with creating for example the
structure you have in your project on separate instances in bots
project and later, just switch the current execution nodes to another
master server and slowly reschedule all running tasks to new execution
nodes configured as you think it's best. That way all bot operators
wouldn't even notice the difference! They wouldn't need to anything at
all.

Or if that would be too much work, why not just to change the
configuration of current bots instances - I don't believe they are so
broken as you think. Since we launched bnr boxes - nobody who was
running anything there yet complained about any issues.

On Thu, Mar 14, 2013 at 2:52 PM, Marc A. Pelletier <marc at uberbox.org> wrote:
> On 03/14/2013 09:37 AM, Petr Bena wrote:
>> BTW @Coren - why do we even need to move current bots project content
>> to another project - wouldn't it be easier for bots operator just to
>> modify the current project to what you proposed, so that they don't
>> need to care about anything?
>
> The primary reason is to make 100% certain that no disruption is
> occurring on the tools that /do/ run on bots.  Especially at first,
> while I'm still shuffling configuration around and tweaking stuff, the
> landscape is moving fast and I wouldn't want to break something that is
> currently working.
>
> -- Marc
>
>
> _______________________________________________
> Labs-l mailing list
> Labs-l at lists.wikimedia.org
> https://lists.wikimedia.org/mailman/listinfo/labs-l