Packet loss

jason_lundejason_lunde Member Posts: 567
Guys,
Just a quick question. Without hindering a production network...what is the best way to see if my 4948 switch is the cause of my packet loss? Any ideas?

Ive already use the show interfaces command...

Comments

  • gojericho0gojericho0 Member Posts: 1,059 ■■■□□□□□□□
    Can you give us more background info into your problem?

    is it packet loss (or connectivity) to a specific device or is there performance issues on the LAN that may be attributed to packet loss?

    Is it occuring all the time?

    What is the output of your show interface?
  • kryollakryolla Member Posts: 785
    look at your buffers but as GoJericho said we need more info
    Studying for CCIE and drinking Home Brew
  • malcyboodmalcybood Member Posts: 900 ■■■□□□□□□□
    miror the port you think is dropping packets and run a packet sniffer to look for retransmissions.

    Also check the port settings that they are not auto-negotiating to half duplex and there are no duplex mismatches.

    Depends what the issue is though as others have said.

    show interface is the best place to start and look for CRC errors, collisions etc.

    Also if you have any qos policies check your policy traffic isn't exceeding allocated bandwidth if appliccable.
  • jason_lundejason_lunde Member Posts: 567
    Sorry, I should have given some more background information. Basically we have a DC that is using two nics in a hp nic team configuration. The nics are set up to use transmit load balancing, and transmit out of both nics, while only receiving data into 1 (each nic was plugged into different 4948's). Show interfaces gave me nothing...and the physical nic stats on the DC gave me no errors as well. So I initially assumed that some nodes were using the mac address of the transmit only nic...and therefore the packets were being dropped. But after remembering the way that MAC addresses were placedinto ethernet frames i realized that the core was the next hop (switched vlans to get to server). So I checked the arp table in the core and it had the correct mac address for that server (the mac address corresponding to the receiving nic). So, in the end it turned out to be a bad nic....(realized this by disabling individual nics and letting the other nic take over). So thanks for the help guys, and sorry about the lack of a good description; I was in a hurry when i initially posted.
Sign In or Register to comment.