topic Re: SNMP engine discovery timeout! in Storage Networking

SNMP engine discovery timeout!

Hitesh Thappa — Tue, 23 Nov 2010 08:27:55 GMT

we get the following error while trying to login to the fabric.

"SNMP engine discovery timeout. It could be

---busy newtwork

---no route from null/0.0.0

---10.64.18.61 snmpd is unresponsive"

but sometimes we are able to login. If we keep on trying to login after several attemps we are able to login but we get Timeout(server, client) error on the core switch and cannot do any zonings or check device manager for that switch.

Other switches in the fabric are working fine.

any suggestions, could this be some issue with ethernet card for the switch?

Attaching screenshots for more clarification on the issue.

Re: SNMP engine discovery timeout!

mitchell.white — Tue, 23 Nov 2010 16:12:27 GMT

Did this behavior just start recently? I am having a similar issue that started last week . Working to resolve it with my vendor.

Re: SNMP engine discovery timeout!

Hitesh Thappa — Wed, 24 Nov 2010 06:28:49 GMT

Yes, for us too this started recently. To me its looks like some ethernet card or ethernet cable issue.

wats the vendor's observsation in your case.

For us like today its working fine with no issues..but noone knows when it will be unavailable howvever we have seen the trend that around 10:30 PM CST it goes down and after midnight its back again.

Re: SNMP engine discovery timeout!

mitchell.white — Wed, 24 Nov 2010 13:51:26 GMT

I haven't received anything back from the vendor (they have a case opened with Cisco). I was asked to provide the FM logs to them.

If you look at the fms_snmp.log file, it is full of these messages:

2010.11.24 02:50:46 WARN [snmp.tcp] TIMEOUT 02:50:46 elapsed=21000 tcp 02:50:25 172.27.xx.xx, 1606970293 BULK u=xxxxx maxRptr=13 clmLicenseFlag....

2010.11.24 02:50:57 WARN [snmp.tcp] Response after timeout (in process packet): 02:50:57 172.27.xx.xx, 1606970293 RESP u=xxxx clmLicenseFlag.....

Of note to mention, I installed Fabric Manager 5.01a on another server and duplicated the timeout behavior. I uninstalled 5.0.1a and installed Fabric Manager 3.3.4a and it works without any of the issues we are seeing.

Re: SNMP engine discovery timeout!

mitchell.white — Mon, 06 Dec 2010 21:03:31 GMT

It looks like we may have discovered the cause for our timeout issues. Fabric Manager was attempting to manage two UCS environments for which the SNMPv3 user was not set. The explanation given was "The UCS authentication failures affect the overall performance of FM due to CPU expensive exception handling".

The timeouts went away once I established the same SNMPv3 user and password on the UCS's that we have set for the seed switches on the fabrics.

Re: SNMP engine discovery timeout!

Hitesh Thappa — Fri, 31 Dec 2010 04:58:16 GMT

Hi MItchell,

Tried your method but still facing the same issue. Moreover when there is timeout on switches they cant be accessed from CLI as well.

This is wat we get in fabric.

NDCCOR31 10.64.18.61 20:00:00:0d:ec:a2:a5:c0 timeout(Server) Cisco DS-C9513 3.3(5)
NDCEXP01 10.64.18.65 20:00:00:0d:ec:a2:3d:40 timeout(Server,Client) Cisco DS-C9513 3.3(5)

Re: SNMP engine discovery timeout!

blueyonderguy — Mon, 03 Jan 2011 04:36:53 GMT

check if this is related

https://supportforums.cisco.com/thread/2050017

Re: SNMP engine discovery timeout!

Kris Vandecruys — Fri, 07 Jan 2011 13:51:13 GMT

While it's not a bad idea to keep up with recent version of FM i don't think that this is currently the situation: if memory serves me well the switch wwns don't match the issue.

This seems more of an IP issue, especially because of this "they cant be accessed from CLI as well."

My suggestion would be to check if you can connect to the switch from a machine in the same subnet as the mgmt ip. If that works, It's most probable a routing/firewall issue