On Tue, 29 Dec 2009, Matthew Melbourne wrote:
Thanks for your reply. It looks like it could be oversubscription of the session path, though I am not completely familiar with the internal architecture of the FWSM. system/xxx# show np all stats | inc Close PKT_CNT: Close indication sent : 1929469548 PKT_CNT: Close indication sent : 946451842 Close Indications : 4197817416
Ok so looks like we even wrapped around, since the sum of the first two is smaller than the third.
Close Notify Errors : 16 How could I determine if this is genuine oversubscription?
That will depend quite a lot on the traffic mix and the configuration, as well as on the available means of telemetry - so unfortunately it's hard to give a one-size-fits-all approach that would fit into a single mail :-)
I'll send you some more questions unicast to understand better what we can do.
Are there any recommendations for mitigating the issue with the current code base?
The "easy" way in this scenario is to not limit the total number of connections - that would at least prevent the blocking that you observe.
kind regards, andrew _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
