Monitoring hosts with unstable connection

[WillemDH]

Hello,

What would be the best way to monitor hosts that have an unstable connection. For example I have to monitor 40 traffic boards, which are connected to the network through gprs, which is very unstable. A board might have 30 - 90 % pl.

I would really need to avoid these boards showing up in open service problems : open host problems all the time. In fact I do not want any of these traffic boards to show up in open service problems. I will have to create a bpi group that says if 3 or more traffic board are having connection problems, I do need an alert.

Thanks for any advice about this setup.

Grtz

Willem

[tmcdonald]

Passive checks probably will be the best course to take. If they try to send and fail due to packet loss, Nagios won't know and won't mark them as Critical. The BPI part will be tricky because they need to be Critical in order for BPI to do its thing, and if they are Critical they will appear as problems. Somewhat of a Catch 22.

[WillemDH]

Hey Trevor,

Thanks for the input, but passive checks are not enough, as the traffic boards are not able to send passive traps. I also need to check connection on a non usual port.

So would there be any other solution to prevent unstable hosts polluting the open service problems view?

Grtz

Willem

[lmiltchev]

Creating a BPI group and playing with the health thresholds would be probably a way to go.

http://assets.nagios.com/downloads/nagi ... _Addon.pdf

[WillemDH]

Ludmill,

Yes, that was the plan. But how do I prevent single traffic boards from showing up in open service problems?

Grtz

[tmcdonald]

I don't think that will be possible. BPI has no mechanism to overwrite what counts as a "real" problem and prevent it from showing as such.

[WillemDH]

So there is no way to prevent hosts from showing up in the open service problems? I can't be the only person using Nagios who needs to monitor hosts with unstable connection?

I really like to keep the open service and hosts problems as clean as possible. I'm monitoring two traffic boards now that keep popping up in the open host problems, but I will have to add 45 more, which will be a nightmare for me...
We only need to be alerted when 3 or more traffic boards are having issues. Do you think my question is worth a feature request? A small checkbox to prevent hosts or services from showing up in open service / host problems would be enough.

Grtz

Willem

[tmcdonald]

So there is no way to prevent hosts from showing up in the open service problems?

Not currently, no.

I can't be the only person using Nagios who needs to monitor hosts with unstable connection?

Most people would mitigate this with an appropriate max check attempts value. It's a bit of a special situation though (90% pl is pretty high) so a perfect solution might not exist right now.

Do you think my question is worth a feature request? A small checkbox to prevent hosts or services from showing up in open service / host problems would be enough.

Honestly I can see this being pretty useful, so I'd say so.

[rajasegar]

Hello,

What would be the best way to monitor hosts that have an unstable connection. For example I have to monitor 40 traffic boards, which are connected to the network through gprs, which is very unstable. A board might have 30 - 90 % pl.

I would really need to avoid these boards showing up in open service problems : open host problems all the time. In fact I do not want any of these traffic boards to show up in open service problems. I will have to create a bpi group that says if 3 or more traffic board are having connection problems, I do need an alert.

Thanks for any advice about this setup.

Grtz

Willem

Create a wrapper script to hold all the individual checks.
Capture how many groups are having connectivity problem.
If more than 3 then exit with error code 2.

You will only have a single check for all the groups.

[WillemDH]

Thanks Rajasegar. Your suggestion seems like the best option at the moment.

Grtz.

Willem