perfSONAR User Q&A and Other Discussion

[Phil Reese <>]

Hey Ivan,

I've been quiet for a day as it seems like we've achieved stability again! 

And you were correct, it had to do with the configuration file.

I had a .conf file which was converted the .json and all the rest.

I noticed some wording change desired on the Dashboard so I simply edited the .json file and restarted all the usual suspects.  Apparently I twiddled something or it otherwise knows that the .json file has been touched.   I went back, made the small edits in the .conf file and pushed it through all the other steps.

My dashboard has been running for two days now and the number of DFOREGROUND processes have stayed right around the original number, ~17+-2.

Fingers crossed that things keep working.

Thank you very very much for your help and patience walking me through the possible problems!

Phil

On 8/8/18 1:28 AM, Garnizov, Ivan (RRZE) wrote:

Hi Phil,

 

It appears to me, you have some MaDDash configuration issue, which results in increased Apache load, which accumulates with time and in the end crashes Apache.

Please start monitoring the usage of memory and cpu on the host and by Apache in attempt to relate to the problem to the issue.

 

The numerous Nagios check failures result in numerous Apache requests and if these timeout, then they put load on the Apache.

If there are Nagios check failures, these should be observed on the GUI as well. Is this true?

 

The simplest example: You have left the “example” MaDDash config lines in the maddash.yaml file.

Please check all the dashboards you may have and seek for “no data” cells.

 

WRT events in “/var/log/httpd/error_log”

Unfortunately the normal operation of Cassandra and Esmond lead to these false positives in the Apache log. Please note these are regular on the hour.

 

Regards,

Ivan Garnizov