Monitoring routers with accurate dependancies

Carroll, Jim P [Contractor] jcarro10 at sprintspectrum.com
Fri Feb 28 17:20:32 CET 2003


I don't recall the details of what happens to the downstream checks if Int0
is down.  But it does seem somewhat pointless to attempt to check something
which is clearly behind a parent.  Having said that, I've never had to
disable notifications; so long as I acknowledge the downed router interface,
I'm good to go.  Once the router interface comes back up, of course
everything downstream to that will be checked (or at least relevant
notifications will resume).  Details escape me, but suffice it to say that
my pager doesn't vibrate until my hip goes numb.

Interesting point about the host vs. service checks.  I haven't really paid
close attention, because the behaviour has been
pleasant/acceptable/desireable.  Yes, if a host is down, you shouldn't be
getting any notifications on services for that host.  It does seem somewhat
pointless to me to be checking services if the host is down, but that could
be a point of logic which doesn't really require tuning, since the
notifications work as desired.

Having said that, I can say from experience that if you have a
servicedependency defined, all services dependent on that one service won't
be checked.  Just to illustrate that point, here's an example:

I've had a (non-critical) host down for more than 4 days 21 hours.  I have
13 services defined for this host.  Of these services, 11 of them are NRPE
checks.  I have servicedependencies defined such that all (but one) of the
NRPE checks are dependent on a trivial NRPE check.  None of the dependent
services have been checked since the depended-on NRPE check failed.  What's
amusing, is one of those NRPE checks still shows green/OK.  :)  (The last
check of this service was performed around the time that the other dependent
service checks were run, as you might expect.)

Anyway, my main point was that the parents directive works as
designed/desired for us here.

jc


> -----Original Message-----
> From: Ian Chilton [mailto:mailinglist at ichilton.co.uk]
> Sent: Friday, February 28, 2003 6:53 AM
> To: Carroll, Jim P [Contractor]
> Cc: nagios-users at lists.sourceforge.net
> Subject: Re: [Nagios-users] Monitoring routers with accurate
> dependancies
> 
> 
> Hello,
> 
> > - if Int1 goes down, you would be notified only of that
> >   - all host and service checks of hosta would wait, 
> pending the resumption
> > of Int1
> > - if Int0 goes down, both Int1 and Int2 (and of course 
> hosta and hostb)
> > would be ignored until the resumption of Int0
> 
> I use the parent feature and my behaviour seems to differ to this.
> 
> If Int1 goes down, everything under it still gets checked, but if
> it's unavilable, instead of getting marked down, it gets marked
> unreachable. I have then turned off notifications on unreachable, so I
> dont get alerted but the web pae shows it as unreachable.
> 
> If a host is down, services still seem to get checked, you 
> just dont get
> notified. i.e - if I block ping's to a machine, it shows as host down,
> but all the services still show as up. If I take services down, they
> show as down, but i dont get notified.
> 
> 
> --ian
> 


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Nagios-users mailing list
Nagios-users at lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nagios-users
::: Please include Nagios version, plugin version (-v) and OS when reporting any issue. 
::: Messages without supporting info will risk being sent to /dev/null





More information about the Users mailing list