The names of maas-django.log and pserv.log are confusing

[Graham Binns]

At the moment, we have three main log files in /var/log/maas:

 - maas.log
 - maas-django.log
 - pserv.log

maas.log is an obvious name. It's where MAAS logs — great! What's maas-django.log? What's pserv.log? Why do I have to care?

The answers are:

 - maas-django.log is where the WSGI container logs when things go wrong.
 - pserv.log is where maas-clusterd and all its services log when Twisted blows up (among other things).

The names are confusing at best and obfuscatory at worst. We should rename them — suggested names are region-debug.log and cluster-debug.log respectively — or even *combine* them into one maas-debug.log.

[Gavin Panella (allenap)]

lp:~allenap/maas/regiond renames them thus:

maas-django.log -> regiond.log
pserv.log -> clusterd.log

That includes packaging changes.

[Julian Edwards (julian-edwards)]

On Wednesday 19 Nov 2014 00:02:38 you wrote:
> lp:~allenap/maas/regiond renames them thus:
>
> maas-django.log -> regiond.log
> pserv.log -> clusterd.log
>
> That includes packaging changes.

I think we actually want these as foo-debug.log as they're not useful to end
users. maas.log is generally intended to be the user-facing one, and these
log names are a little enticing!

Would you consider regiond-debug.log and clusterd-debug.log?

[Gavin Panella (allenap)]

> Would you consider regiond-debug.log and clusterd-debug.log?

Of course :) It's just more work, but not much more.

It would be nice to write maas.log into the database and make it available in the UI and web API. Then grubbing around with nasty files need only happen when things go catastrophically wrong. Which they never do, right?

[Julian Edwards (julian-edwards)]

On Wednesday 19 Nov 2014 14:56:49 you wrote:
> > Would you consider regiond-debug.log and clusterd-debug.log?
>
> Of course :) It's just more work, but not much more.
>
> It would be nice to write maas.log into the database and make it
> available in the UI and web API. Then grubbing around with nasty files
> need only happen when things go catastrophically wrong. Which they never
> do, right?

Easy, it's all syslog, JFDI a syslog client :)