Re: [Nagios-users] Trends Graph not graphing short outage

[Peter B. <be...@pu...>]

On Tue, 14 Oct 2003, Mike McClure wrote:

> Hiya Peter!
>
> > Just installed nagios 1.1.  I'm using it to track and document the high
> > round trip times and packet loss so my business internet provider can fix
> > it.
>
> Sounds like a fine idea.  Hope you have a good SLA from them.

 Heh -- I wouldn't call it an SLA, but telling them that we lose
 connectivity 2-5 times a day should warrant a service call...

> > The ping service registered a warning that lasted about 3-4 minutes, but
> > that warning period does not show up on the graph, even though the graph
> > shows the current accurate time.
> >
> > The history log:
> > [10-14-2003 09:29:49] SERVICE ALERT: nambco-cable;PING;CRITICAL;SOFT;1;CRITICAL -
> > Plugin timed out after 22 seconds
> > [10-14-2003 09:30:49] SERVICE ALERT: nambco-cable;PING;OK;SOFT;2;PING OK - Packet
> > loss = 0%, RTA = 36.01 ms
> > [10-14-2003 09:46:49] SERVICE ALERT: nambco-cable;PING;WARNING;SOFT;1;PING WARNING -
> > Packet loss = 0%, RTA = 124.70 ms
> > [10-14-2003 09:47:49] SERVICE ALERT: nambco-cable;PING;OK;SOFT;2;PING OK - Packet
> > loss = 0%, RTA = 40.86 ms
> > [10-14-2003 16:59:50] SERVICE ALERT: nambco-cable;PING;WARNING;SOFT;1;PING WARNING -
> > Packet loss = 35%, RTA = 45.78 ms
> > [10-14-2003 17:00:49] SERVICE ALERT: nambco-cable;PING;OK;SOFT;2;PING OK - Packet
> > loss = 0%, RTA = 81.64 ms
> >
> > The graph shows all green, with a starting point of (currently) October 13
> > 17:22:36 and an endpoint of October 14 17:22:36.
> >
> > Is this a bug, or is this because I said "include start-up state" or
> > whatever that feature is.
>
> Which graph are you talking about?  Trends, Availability, or Alert Histogram?

 Trends.

> As written, the only one that shows SOFT states is the Alert Histogram.  However,
> you could very easily comment out the 'if(strstr(input_buffer,";SOFT;"))' statements
> in the C code for those CGIs and recompile/reinstall to get the SOFT states in those
> graphs.
>
> Alternatively, you could set the "max_check_attempts" for the PING service to 1.
> That will make it go into a HARD state immediately.  But that will also trigger an
> alert, that may or may not be something you want to happen.
>
> I hate to tell you this, but I don't see anything lasting 3-4 minutes according to
> your logs.

 Ah!  I just finished using Netsaint, which DID graph SOFT states (I
 believe); knowing that SOFT states don't get graphed in the Trends graph
 would answer the question.

 Maybe I really just need to stick with the check_attempts at 3 (default)
 and hope the outage lasts more than 1 minute.  I guess that makes sense.
 SOFT states are logged but not graphed, except on the Alert Histogram.
 Got it -- thanks for the verbose explination!

> Mmmmm..   Grape soda and ice cream...

 Heheh... It's a Michigan thing I think... good stuff.

Beckman
---------------------------------------------------------------------------
Peter Beckman                                                  Internet Guy
be...@pu...                             http://www.purplecow.com/
---------------------------------------------------------------------------