02-27-2014 04:06 AM - edited 03-07-2019 06:26 PM
We have a stack of 3 2960S's switches connected to a Cisco 3750X via a 4 port copper Etherchannel, intermittently the users are losing connection to the internet and with a host connected to the Master switch running a continuos ping to the 3750X pings are lost, we have replaced the Master switch and both GLC-T's. We also split the stack and had a single 1Gb connection to each switch from the core, we ran continuos pings from hosts on each switch and the only one to drop pings was the one on the Master switch. Versions and configs below, any thoughts ?
Version 3750X - 15.0(2) SE4
Version 2960S - 15.0(2) SE5
Port-Channel Config 3750X in attached file called 3750x
Port-Channel Config 2960S in attached file called 2960s
Also auto QoS has been enabled on both 3750X and 2960S.
Nothing appears in the logs on both switches.
02-27-2014 07:22 AM
Hi Will,
Would it be possible to post the output of the ping response to the 3750 when you connect to the master switch. Few questions taht may help us troubleshoot this.
What is the load balance method you are using for the etherchannel ?
Is it only the pings that are failing or is the user traffic affected to when using the master switch ?
Can you post the source(PC) and destination ( 3750) MAC and IP addresses ?
HTH
Regards
Umesh
02-27-2014 11:01 AM
Umesh,
Thanks for the reply, I will be doing more testing tomorrow, load balancing method is the default which I beleive is source-mac. It's not only pings but all users are unable to reach the internet, or local servers.
Regards Will
02-27-2014 01:39 PM
Do a continuous ping from client PC to the IP Default Gateway of their IP Address. See if that drops out.
02-28-2014 08:33 AM
Hi Will,
While you test can you see you could do these tests as well.
1> First as Leo said try pinging the default gatewat from a PC connected to the master.
2> Secondly can you check on the 3750 if you are able to learn the mac addresses of the PC's connected to the master on the 3750 stack?
3> Can you try pinging a PC on the member from a PC on the master.
HTH
Regards
Umesh
02-28-2014 01:17 PM
Leo/Umesh,
Again thanks for the replies, I did a continous ping from a PC connected to the switch stack to an address on the Internet, I also had a continous ping going from another client on a different switch stack to the same internet address. The PC connected to the suspect switch stack dropped about 15 pings but the other PC was fine. I've now set up two syslog servers with the stack logging to one and the core to the other, also I've changed the stack master to be the middle switch which has no up links connected to it. Customer will start monitoring on Monday.
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide