Rx-No-Pkt-Buff - oversubscription WS-X4548-GB-RJ45

[Jeroen Huysmans]

Hi,

I'm trying to understand the counter Rx-No-Pkt-Buff on module WS-X4548-GB-RJ45. We have a Supervisor Engine IV in this WS-C4506 chassis.

I know this card has 6 portgroups in which members share 1Gbps throughput.

I'm seeing inpunt errors on 2 interfaces which correspond 1:1 with the Rx-No-Pkt-Buff counter. The device connected to these 2 ports is a Cisco 3020 switch in an HP Blade enclosure.

Gi6/43

     101 input errors, 0 CRC, 0 frame, 0 overrun, 0 ignored

Port         Rx-No-Pkt-Buff     RxPauseFrames    TxPauseFrames   PauseFramesDrop
Gi6/43                  101                 0                0                 0

Gi6/44

     78 input errors, 0 CRC, 0 frame, 0 overrun, 0 ignored

Port         Rx-No-Pkt-Buff     RxPauseFrames    TxPauseFrames   PauseFramesDrop
Gi6/44                   78                 0                0                 0

My guess: the port-group is oversubscribed...

However, only 2 ports in this port-group are in use: 6/43 and 6/44. Both are configured as etherchannel members (which is probably not a good idea because they share the same asic).

According to my graphing (cacti, 1 minute polling) both ports do not generate a total of 1Gbps. CLI output confirms the graphs.

How can I troubleshoot this further? Why is the Rx-No-Pkt-Buff increasing while we don't exceed 1Gbps of bandwidth in this group?

I'm aware this linecard is "probably" not the best to use to connect a Blade enclosure ("designed for gigabit to the desktop"), but I'm just trying to understand why we see this errors...

regards,

Jeroen

[glen.grant]

  I think you have a pretty good handle on it .  Seeing that you basically have a whole server farm (bladecenter) plugged into those ports it's probably being overrun .  I would change ports in your port channel  or try running a 4 port etherchannel  with 2 of the ports on a completely different card. If those statistics are on ports that have never had the stats cleared that really isn't a lot and is probably not a reason for concern.  Clear the ports and watch it for a week , if it's just a few I probably wouldn't worry about it and just occasionally revisit it specially if you are getting slow performance complaints.