Re: [hobbit] A few questions before moving from bbgen to Hobbit

[Frédéric Mangeant <frederic.mangeant (at) steria.com>]

> > I'm preparing the "big move" from bbgen to Hobbit on a large production 
> > setup, and I have a few questions :
> >
> > - as I'm running Cacti to graph my servers, is there a way to remove the 
> > "trends" colum for all hosts ?
> > I've disabled the rrdstatus and rrddata tasks, added TEST2RRD="" and 
> > GRAPHS="" but the "trends" column still show up, empty
> >    
>
> You can do it on a per-host basis with the "notrends" setting in
> bb-hosts, and this can be set as a default using the ".default."
> host entry in bb-hosts.
>
> You cannot do it globally, except by tweaking the code in hobbitd.c
> that adds this pseudo status.
>  

Thanks, I've added a .default. host with "notrends badconn:..." and the 
trends column disapeared.


> > - in the Administration -> Enable/disable page, is there a way not to 
> > show the reason of the disabled tests ?
> > With a few hundred disabled tests, the page gets difficult to read
> >    
>
> No, not without a code change.
>
> Note that the enable/disable page also works on subpages, and in that
> case it only shows those hosts that appear on the subpage. It might
> still be many, but perhaps it helps.
>  

I didn't notice that, it's fine.

> > - in the "info" page of each host, would it be possible to show tests 
> > with no alert rules ?
> > With bbgen they were displayed as "Ex.Services"
> >    
>
> It would be fairly easy to add a line at the bottom saying "No alerts
> defined for: disk, msgs, conn" ... is that what you're after ?
>  

Exactly, just to be sure that I don't miss alert rules if  EXSERVICE  is 
used.


> > - do you plan to add file inclusions to hobbit-alerts.cfg, like with 
> > bb-hosts ?
> > We're several people managing the BB/Hobbit server, and this would be 
> > very useful.
> >    
>
> It's on my list of pending cleanups. It needs a bit of work because
> there's some logic currently that avoids re-loading the configuration
> unless it's changed - and that currently works by looking at the time-
> stamp of the file. This obviously has to change, and I haven't yet
> decided what the best solution is.
>  

Great.

--

Frédéric Mangeant

Steria EDC Sophia-Antipolis