Hi All,
I have been testing the new WSMan based Zenpack (http://wiki.zenoss.org/ZenPack:Microsoft_Windows) and have found it to be a very powerful framework, so many thakns for all the hard work which has gone into getting it this far. However on all the machines I have been testing against it stops collecting data randomly and then starts collceting again the next time the configuration is updated :-( I'm hoping that someone which better skills that me or who has been involved in the development can shed some light on this issue and help find get to the bottom of it?
I've turned on debug logging for zenpython (the collector daemon that does the collection) and attached an extract of the logs, showing successful collection, which then just stops with no errors. I've checked with netstat at both ends and when collcetion is working a WSMan connection is maintained, but once it has stopped there is no WSMan connection and no attempt is made to create one :-( There is nothing in any of the other zenoss logs and nothing in the windows event logs to indicate an error.
Looking at the source code (PerfmonDataSource.py, lines 178-205), I can see that the collection starts a remote shell and runs the "get-counter" Powershell command. It appears that this is configured to run for a fix times of 10 minutes and report samples at the interval configured as the cycle time:
...
if self.cycling:
self.sample_interval = dsconf0.cycletime
self.max_samples = 600 / self.sample_interval
...
'get-counter -ea silentlycontinue '
'-SampleInterval {SampleInterval} -MaxSamples {MaxSamples} '
...
It seems to me that what is happening is that the get-counter command is finishing (after it's 10minute run) but that the collector task is not starting a new get-counter command as it normally would (it seems like it might still be waiting for more data from WSMan, even though the conncetion has been closed).
Has anyone else experienced the same issue? Or is anyone from the development team reads this, perhaps you could exlpain the logic behind having a fixed run time of 10 minutes for the get-counter command? In futuer I'd like to have some monitoring templates which only collect data hourly but these will error with teh current code as far as I can see?
Thanks,
Nick.
