TECHZEN Zenoss User Community ARCHIVE  

Threshold aren't respected in Zenoss 3.1.0

Subject: Threshold aren't respected in Zenoss 3.1.0
Author: Anand Asokan
Posted: 2017-05-29 22:24

Hey Zenoss Gurus,

We made some changes in our zenoss system for a couple of servers to monitor one filesystems ONLY if it crossed 90% utilizations.

We followed steps depicted in this wiki especially the last two parts for monitoring individual FS.

However, alerts are still being triggered even if utilization < 90%.  
How is this even happening? We are still newbie to Zenoss so we wanted to post this question to the experts out there.

Zenoss 3.1.0
Linux
Python 2.6.2


Thanks!
A

------------------------------
Anand Asokan
Unix Admin
ABB
------------------------------


Subject: RE: Threshold aren't respected in Zenoss 3.1.0
Author: Jane Curry
Posted: 2017-05-30 09:06

Couple of thoughts here.  If the sampled value is very close to 90%, I have seen a threshold still fire - either because of the way  that the data is actually collected or because of rounding errors in the calculation. 

If this is obviously not the case, then navigate to the device that your getting the event for and then select the file system that is causing the event.  You should see data for the filesystem in the main part of the window - Total, used, free and percent.  Does this data exist and is it valid?

Also check that this filesystem does, indeed, have a local template - select the filesystem and change the Display dropdown to Templates - the description should show the path to the device and filesystem.

You might also look at the detail of the threshold event - double-click it in the event console and scroll to the bottom to the "Event Details" section.  You should see values for Current and max / min there - are they "sensible".

Do ensure that you haven't changed the name of the template being used.  Component templates get bound by "magic" - the template name must exactly match the object name that is defined in the Zenoss code.  Some people try to "play safe" and take a copy of a template and then rename it; that's usually fine for a template that works on a device but it will break things if you do this for a component template like FileSystem or ethernetCsmacd (for interface components).

Cheers,
Jane

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


Subject: RE: Threshold aren't respected in Zenoss 3.1.0
Author: Anand Asokan
Posted: 2017-05-30 21:55

Fantastic Reply Jane, going to check on it and feedback.

------------------------------
Anand Asokan
Unix Admin
ABB
------------------------------


Subject: RE: Threshold aren't respected in Zenoss 3.1.0
Author: Anand Asokan
Posted: 2017-06-08 23:20

OK checked all suggestions, but still alerts are fired. Suspect now its a bug! BTW, do we need to push changes after creating a local copy? Thats one thing we never done yet.

-Anand

------------------------------
Anand Asokan
Unix Admin
ABB
------------------------------


Subject: RE: Threshold aren't respected in Zenoss 3.1.0
Author: Jane Curry
Posted: 2017-06-12 07:36

Even with Zenoss 3.1 I doubt that you need to actively Push Changes.  By now, the Config Cycle time will be up so the changes should certainly have been pushed (in later - and I'm not quite sure exactly what I mean by "later" here - any changes automatically get pushed when they are saved).

Any chance you could post some screenshots with the event detail and the template definition (if you would rather not post publicly, happy for you to email them to me).

Cheers,
Jane

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


Subject: RE: Threshold aren't respected in Zenoss 3.1.0
Author: Anand Asokan
Posted: 2017-07-12 21:07

Jane, sincere apology for not replying lately, just like any sysadmins life I got pulled into another project and issues.

Till lately, I've found a little trick where you can also opt to ignore the whole directory altogether from being monitored. Its referenced here: the part that says to edit zFileSystemMapIgnoreNames.

Another option that I looked into personally was just disable monitoring from components in zenoss devices itself. 

I did the latter and have to wait till this weekend if it really works.

Can I still keep you posted?

-Anand

------------------------------
Anand Asokan
Unix Admin
ABB
------------------------------


Subject: RE: Threshold aren't respected in Zenoss 3.1.0
Author: Jane Curry
Posted: 2017-07-13 05:29

Please do.
Your trick for unmonitoring components is quite a good one.  The problem comes if you have LOTS of components - you don't want to manually manage/unmanage them all (and there were some bugs around in older code around the component manage / unmanage flag in the GUI).  I wrote a little script some time back that helped with this for specific types of components - feel free to pull it from  httpshttps://gist.github.com/jcurry/eddbcf819d1f8c17848431cb577a7152://gist.github.com/jcurry/eddbcf819d1f8c17848431cb577a7152 .

Cheers,
Jane

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


Subject: RE: Threshold aren't respected in Zenoss 3.1.0
Author: Anand Asokan
Posted: 2017-08-01 03:06

Hey Jane, that method worked like a charm, but its not a permanent fix but hell of a good workaround! :)
Its been couple of weekends or more and no more alerts for that one dir and less complaints from users. Yup they agreed to have that totally unmonitored too.


Anyways, I'm going to have a look at your git repo now. Thanks!!! 


-A

------------------------------
Anand Asokan
Unix Admin
ABB
------------------------------


< Previous
Save Upgrade Procedure
  Next
Threshold aren't respected in Zenoss 3.1.0
>