CHECK_NRPE: Socket timeout after 30 seconds.

[srikanth.kallu]

Hi

I had this error "CHECK_NRPE: Socket timeout after 30 seconds." on one of my linux server and recoverd itself in about 15 minutes. We thought everything is ok but leter found that nobody can login to the server.

We had to powercycle the server to bring it back online. and later found from the logs that there is an I/O error and kernel failed error

But i need to know how/why did nagios show everything is fine and what is the fix for it

Please advice

Thanks,
Srikanth

[Box293]

What was the check being performed?

[srikanth.kallu]

I was checking all the file systems, CPU, Load, Users, number of procees and open files for all of them i got this error and recovered by itself

[Box293]

We thought everything is ok but leter found that nobody can login to the server.

Is there a way to check that the server is allowing logins?

[srikanth.kallu]

No, but as because showed in the logs it got a kernel failed error and I/O error i was expecting load or CPU or something should have shown critical . but all showed OK

[Box293]

Perhaps you need to be monitoring the logs.

In terms of load and CPU checks they all returned results within the thresholds. You can't expect those checks to tell you about a kernel failed error as that is not their purpose.

[srikanth.kallu]

Is there any way we can setup to monitor syslog ?

We are planning to build a centralised syslog server so that all servers sends logs to that. Is there any way we can monitor those logs ?

Do anybody you know do this ?

[Box293]

There are plugins available and different methods to do this however it can get quite complicated.

If you are planning on building a centralised syslog server, we have a presentation next week at the Nagios World Conference that covers this topic.

http://www.nagios.com/events/nagiosworl ... -Wilkerson

I cannot say much else right now however after Tuesday I can go into more details.

[srikanth.kallu]

do you mean to say there are plugins that exist now are complicate and there is a simple way of building centralised logging which will be presented next week ?

Is it a webinar?

[Box293]

do you mean to say there are plugins that exist now are complicate and there is a simple way of building centralised logging which will be presented next week ?

There are plugins that monitor logs however active checks with Nagios to monitor logs don't always work as expected and it can be very complicated and time consuming to setup.

Is it a webinar?

No you have to attend the conference to see the talk. I cannot say any more than what I have said and provided a link for.