SNMP monitoring gaps

  • 1
  • Question
  • Updated 3 years ago
We are using LibreNMS to provide an additional view into our network, but it seems to give a lot of false positives when reporting on Aerohive access points.  As you can see in the graph below, there are gaps where we aren't receiving info from the access point, but at the same time, I can ping the access points, I can see that clients are connected via 'sh station' at the cli, and I can ping the clients connected to the access point.  I asked a question about SNMP a couple of years ago, but are we now at the point where we should be able to reliably pull SNMP information from our access points?

Photo of Van Jones

Van Jones

  • 75 Posts
  • 4 Reply Likes

Posted 3 years ago

  • 1
Photo of Mike Kouri

Mike Kouri, Official Rep

  • 1030 Posts
  • 271 Reply Likes
Van,
What is the age of the devices you are querying? If they are older generation devices like the AP121/141, AP170, etc. then it is not unlikely that they are busy and we will always give precedence to handling user traffic over satisfying management queries.
Photo of Van Jones

Van Jones

  • 75 Posts
  • 4 Reply Likes
All we have are AP330 and AP230 access points. Should I expect the same behavior that you described from the 330 and 230 access points?
Photo of Mike Kouri

Mike Kouri, Official Rep

  • 1030 Posts
  • 271 Reply Likes
Van,
The underlying logic within HiveOS is the same, so it is possible. The AP330 and AP230 have both more DRAM and more powerful CPUs, so they would have to be much busier than an AP121/141 would be before they started exhibiting the same symptoms. 

Are your users complaining about connectivity issues when these 'dead SNMP' intervals happen? Since you didn't say so, I assume not.

It is possible that you're just encountering a bug in our code. I looked a little more closely at the graphic you supplied, and noticed that the 'dead' periods seem to be immediately or shortly after a block of 'busyness". Conceivably there's some buffer exhaustion happening, and it takes a little bit for it to recover and become responsive again.

One of the key goals in our upcoming HiveOS 6.5r3 release was finding and fixing all known occurrences of Out-Of-Memory conditions, and to improve the recovery times for those which absolutely cannot be avoided.
Photo of Van Jones

Van Jones

  • 75 Posts
  • 4 Reply Likes
We are in a higher ed campus environment and our students rarely complain unless it affects their gaming experience. :)  

I do have a question about HiveOS releases, but I will start a new thread since I would be off topic asking it here.