PQM is too queued

[Martin Pool]

On 12 Feb 2007, Aaron Bentley <aaron.bentley at utoronto.ca> wrote:
> John Arbash Meinel wrote:
> >>>Stick me behind other bzr requests-- that's sensible.  Waiting 2.5 hours
> >>>for an unrelated request to complete is not.
> 
> > Well, if that other project is Launchpad (which it typically is), I've
> > timed it as each submission taking approx 1hr.
> > 
> > So if you are stuck behind 5 submissions, you are talking about waiting
> > 5 hours, not 2.5.
> 
> So far, the queue has not moved since I submitted my request.  So the
> "Now playing" request has been playing for 2.75 hours, at least.  I have
> no way of knowing how long exactly.
> 
> If we count the remaining 5 requests as 1 hour, then we're looking at
> 7.75.  It's also possible that the remaining 5 will also take 2.75+
> hours each.  I hope not.
> 

> > I've mentioned it too, though my feeling was it may just be best to have
> > a separate pqm instance. Maybe running on the same machine, maybe just
> > running on a less powerful machine, since bzr doesn't take as long as
> > Launchpad to test.

It can rarely happen that pqm is in fact just jammed.  I think this once
happened because the box was rebooted and something didn't restart.  

Robert said that 

<lifeless> it was disabled overnight by stub doing maintenance
 its now running at regular speed

> Yes, some kind of separation would be nice, even if it meant using a
> slower box.

I don't see why a separate machine would be needed, maybe just a
separate email address for our merges going to a different pqm
configuration.

-- 
Martin