Collector sometimes fails to retrieve the information from a WLC with 900 APs.

[victor.vasoler]

Hello,

I hope you could help me to clarify the following doubt, please.

Once a week we collect around 6000+ devices in the collector, which represent around 14000 chassis in total (taking into consideration routers, stacks of switches, WAAS, WLC, APs, etc)

I have noticed that sometimes the collector fails to retrieve the information from a WLC in Asia wich contains around 900 APs (sometimes it works fine and some others the collector is not able to retrieve the information).

I believe that this could be due to the quantity of information that it retrieves from that WLC in particular. Is there a timer that could be exceeded during this operation?

It could also be related with latency from the collector located in US to the WLC located in ASIA.

Is there a configuration or something that I could change in order to avoid this issue?
Everytime that the collector fails to retrieve the info from that WLC, our inventory is reduced by 900 devices, which represents an important fluctuation in our reports.

I hope you can assist me about how to solve this issue or at least how to reduce the risk for this to happen.

Thank you!

Victor Vazquez

[CC Cisco · Accepted solution]

Hello Victor,

Please try increasing the inventory global SNMP timeout.

This can be accessed via Settings -> Inventory Settings -> Global Timeouts.

Regards,

Cesar

[victor.vasoler]

Hello,

I think this is exactly what I need.

At this point we are only using SNMPv3 to collect the data from all our devices.

I see that the default timeout is "5000ms" and the retry count is "2".

Is a good timeout "10000ms"? (just as it is for other protocols like SSH and TELNET).

Or would you recommend a different value here?

I guess the retry count can remain as "2" for this particular case, correct?

Please let me know your comments,

Thank you!

Victor Vazquez

[CC Cisco]

Hello Victor,

A value of 10000ms should be a good starting point.
I’ve seen instances where the timeout has to be increased up to about 50000ms, so it won’t hurt to try with 10k.

The best way to determine the best value is to run the inventory and see if all data is captured within the specified time window.

Regards,
Cesar