perfSONAR User Q&A and Other Discussion

[Mark Feit <>]

Kate Adams writes:

 

Hello All,

 

I'm trying to figure out why some hosts are not green on the maddash even though pscheduler can run tests to and from the one of the hosts. 

 

In the process of reading the troubleshooting guide, I came across the git page, https://github.com/perfsonar/pscheduler/wiki/Error-Messages

 

I found this error in /var/log/pscheduler/pscheduler.log

 

Mar 10 07:40:24 speedy journal: runner ERROR    32056995: Failed to store run: Invalid transition between states (7 to 5).

 

I don't know if it has anything to do with our issue, but just in case it helps the developers, I'm passing it along.

 

That message is the result of the runner failing an internal check that makes sure what it’s doing is sane.  In this case, the run went more than 30 seconds beyond the time a result was expected, was declared overdue and then the runner came back with a result.  That isn’t supposed to happen, and when it does is an indication that the system is severely loaded down, a plugin isn’t policing its processes or the clock was adjusted forward far enough to trigger this.  I don’t see any harm in a result being reported late, so I’ll adjust the table of what’s considered sane to include that case.

 

Are you seeing many of these or just a few?

 

By the way, the large number in the error message is a running count of the number of measurements that have been taken since the database was first installed.  32 million’s not too shabby.

 

--Mark