Header Only - DO NOT REMOVE - Extreme Networks

C5210 Randomly reporting SNMP Alarms


Userlevel 2
Hi All,

As the title says, one of our C5210 controllers keeps reporting SNMP Alarms to EMC.

Sometimes as down (although I can access the GUI) and sometimes as UP.

I know the device is reachable as it will also report the uptime which will report the current time since last reboot, in this case currently 21 days.

It's getting a little annoying. I have disabled/enabled SNMP. rebooted in between. Tried different versions of snmp etc. Still does it.

It has logged SNMP agent crashes although it doesn't do this every time and hasn't done so in a while, its not inline with the alarms anyway.

will the CLI provide me with more information than the log. Do we think the controller actually has an issue?

We're running 10.31.003.0012.

Any help is appreciated.

Thanks
Ian

5 replies

Userlevel 6
Hi Ian

This is only happening to 1 controller of a pair right?

In the first instance you could try to set that devices timeout/retry in EMC to something a bit longer (default is inherited globally but you can configure per device from 5 secs timeout/3 retries.)

Global is set in Admin>SNMP
Per device is set Network>Devices>Select Device>Menu>Edit



We'd need a case opening and the "tech support all" sending to look at the SNMP crashes.

-Gareth
Userlevel 2
Hi Gareth,

Yes just the one controller in the pair.

I had to reboot the controller to fix it, restarting the SNMP processes did not solve the issue.

I'll raise it with our SE partner and take it from there.
Userlevel 2
Hello Ian,
we are having the exact same problem with our 2 C5210 controllers, once my controllers report down they do not come back up even after rediscovering. However your description is more succinct than mine. We have tried all of the things you have with the same results.
What is your configuration in regards to the physical ports? Is the port that you use to manage the controller (allow management traffic) on the same subnet as the EMC?
In my configuration it is not on the same subnet but it is on a directly connected network. Other devices on that subnet have no problem connecting and talking to the EMC.
One suggestion we had from GTAC was to reduce the number of "physical" ports to 1 on the controller. I deleted all the other physical ports to just the 1 I am using for the APs and management. This seemed to work for a while but then the controllers dropped off again and I could not re-establish communications.
I have connected the management port directly to the same subnet as the EMC. This has provided a stable connection to the EMC console but device collection statistics for OneView does not appear to work. The controllers are always saying down on the OneView dashboards however we do have our AP alarms back which is helpful.
I do not like this work around as it is not working completely and I do not like using the management port as the documentation says not to use it.
I have a case open with GTAC and my own thread on the hub (Is it possible to direct SNMP traffic to a specific physical port on a EWC C5210). I will follow your topic to see if you have any updates and let you know if anything is found from my own inquiries.

Chris
Userlevel 2
Hi Chris,

We use one 10Gb port for the management (ESA2) and tag all vlans to the port. (most VNS are tunneled back to the controller) the port links directly into the core. same for the other controller.

The management subnet is a small /28 subnet with the other controller in it.

the EMC server is on another subnet but routed within the LAN.

There is a simple default route to the management subnet gateway for all traffic. Nothing too fancy here.

Primary controller hasn't had any issues. suspect the other with the issue actually has a problem but as we run all APs to the primary controller as local APs we don't really see any traffic on this problematic controller to determine if any other issues.
Userlevel 2
Hello Ian,

I removed the management port connections and both controllers are still talking to NetSight however one of them still shows down in OneView.
I will update if they go down again.

Reply