6513 performance question

[ciscoforum]

When I type show counters 1/11 I see a"ifInDiscards=2040037" this is probably accumulated more than half year. But does it still indicate some problem on this particular port? We have 1 file cluter which has two servers, one in 6513A port 1/11,the other one on 6513B port 1/11. The server on switch B respond very quick, but the server on 6513A very slow when user need to access the files. Does this show command tell me sth? Anything else I should look at? Thanks

[gpulos]

after you examine and note the counters of discards, perform a 'clear counters' command. this will set the counters back to zero and allow you to do a 'show counters' and see if the discards are still incrementing.

if they are not, then that would be fine. if they are, you'll need to verify why of course...(many things could be a cause)

1) check speed/duplex on server & switch

2) replace the patch cable from server to switch (seen this cause major drops)

3) verify any hidden cabling between the server/switch is OK. (seen this cause major drops)

4) check the switch configuration; ACLs, QoS, routing

5) verify tcp stack on server

it is possible there is another source of the symptom

[Roberto Salazar]

The "ifInDiscards" mean that on the port you are seeing this increments, is you have packets destine to the vlan that is blocking on this port. The port 1/11 is most likely trunked. For example on 6513A port 1/11 is trunk and it has vlan 10 and vlan 20 not in forwarding. While on the 6513B port connecting to port 1/11 of 6513A have all the vlans forwarding. The packets (broadcast/unicast flood/mcast) that is recieved on port 1/11 of 6513 will be counted as "ifInDiscards", it does not mean anything is wrong. You can stop the counter increment by having both ports match the forwarding vlans. The latency issue on the server is another thing but you might be lookng at the wrong counter and it will throw you in a loop. Here is how you can verify if my analysis is correct>

on 6513 A:

show trunk >> look at the last section of the output titled "Vlans in spanning tree forwarding state and not pruned" compared the vlans with the trunk port going to this port from the other switch, you will find that there are some vlans blocking on this port while they are forwarding on the other switch.

Please rate helpful posts.

[ciscoforum]

Thanks for your reply. But both ports are configured as acccess port. I verify this in couple ways, configuration, show trunk, show port etc. I have a feeling it's caused by flowcontrol. The server currently disabled all the flowcontrol though 6513 has flowcontrol send desired. I will turn on receive flowcontrol on server and check the counters again.