TECHZEN Zenoss User Community ARCHIVE  

threshold of zenmodeler cycle time exceeded

Subject: threshold of zenmodeler cycle time exceeded
Author: Paul Giordano
Posted: 2018-09-18 10:23

So I'm receiving an event
Threshold of zenmodeler cycle time exceeded: current value 40866.200768

I notice in the conf there's a "cycletime" setting currently set to "none" (the default). Not sure what that does. What can I look at to figure out what's going on? Where are the logs at? The kibana logs are a bit difficult to navigate, but some pointers here would help as well.

------------------------------
Paul Giordano
Senior Systems Engineer
Zethcon Corporation
------------------------------


Subject: RE: threshold of zenmodeler cycle time exceeded
Author: Jane Curry
Posted: 2018-09-21 05:24

So it is telling you that the zenmodeler cycle is taking 40866 seconds to complete - 11+ hours????.  If you have a huge environment then this might just be justified - more likely that something is not quite right and it shouldn't be taking that long.  Look at zenmodeler.log which should confirm the timing - look for the last line of the cycle which will be something like:

2018-09-09 08:48:32,420 INFO zen.ZenModeler: Scan time: 187.97 seconds

You might also look for timeouts in this log.  If you have lots of devices with long individual timeouts and retries then it can seriously extend the modeler cycle.  You should maybe be looking at using collectors / more collectors to divide up the load.

I find the Kibana stuff opaque so I would connect to the zenmodeler container and then the zenmodeler.log file should be under /opt/zenoss/log.

Have a look at https://support.zenoss.com/hc/en-us/articles/215003503-How-to-Edit-the-zenmodeler-File-to-Configure-Model-Scheduling-in-Zenoss-5-x  .

Cheers,
Jane

------------------------------
Jane Curry
Skills 1st United Kingdom
jane.curry@skills-1st.co.uk
------------------------------

Subject: RE: threshold of zenmodeler cycle time exceeded
Author: Luke Lofgren
Posted: 2018-09-21 07:52

Using Control Center you might also benefit from editing /opt/zenoss/etc/zenmodeler.conf and adjusting this setting:

# Number of devices to collect from in
# parallel, default: 1
parallel 5

For our main collector that does have a large number of instances we run 5 modeler containers, and have each container doing 5 parallel modeling "jobs" as specified above. This means we're actually modeling 25 things at once. This reduces our modeling time to a few hours -- we'd be running longer than you indicated if we left the defaults of 1 container and 1 "parallel". 

Note that you will increase load on ZenHub and other containers the more simultaneous modeling you are doing.

Hope this is helpful. 



------------------------------
Luke Lofgren
Infrastructure Architect
Acxiom Corporation (home based associate)
Waterford PA
------------------------------

Subject: RE: threshold of zenmodeler cycle time exceeded
Author: Paul Giordano
Posted: 2018-09-25 10:32

Thanks Jane and Luke, We're only running 420 servers and thought that the modeler running so high was suspect. I set the modeler to start at midnight, and use 5 threads - we'll see what happens.

------------------------------
Paul Giordano
Senior Systems Engineer
Zethcon Corporation
------------------------------


< Previous
Backing up Analytics
  Next
Zenoss 6.1.1 graphs show no data, zenhub and MetricShipper failing some health c ...
>