4948-10GE sudden intermittent connections

[dan.letkeman]

Hello,

I have a pair of 4948-10GE switches that I use for iSCSI traffic.  They are connected together with 2x 10G gbic's.  All of the sudden one of the switches was having problems and sort of cutting in and out, thus causing interfaces to go up and down rather quickly.  Best practice for the san setup is to turn off spanning tree, which makes sense, but in this particular case this caused issues because of the interfaces connecting and reconnecting rapidly and sort of keeping the connections alive.

*Edit, I didn't disable spanning tree on the enter vlan, but I did enable portfast on all of the ports facing the SAN and Hosts.

The onsite tech just landed up turnning off one of the switches to solve the issue as everything was connected in a redundant fashion, but now I don't have any logs that show what the problem was.  I have then since added syslog exports to the switch, but I was wondering if there is any other way to see what happended?

The only thing I can go by is that an adjacent switch had an EIGRP neighbor relationship with it and I could see that is was going up and down.  I am using the latest IOS available for these switches.

Thanks,

Dan.

[Gabriel Hill]

Hello Dan,

I have never heard of any recommendations in a SAN environment to disable spanning-tree, it actually gives me goose bumps thinking about that. Is spanning-tree disable just for the SAN vlans or for all vlans? Your description is hard to determine what may have gone wrong, but with spanning-tree disabled, a loop would seem suspicious. Are the 2x 10G interconnects between the 4948 in a port-channel?

-Gabriel

[dan.letkeman]

Gabriel,

I guess I missed on line in my explanaition. I meant to add that I didn't disable spanning tree, but I did enable portfast on the ports facing the nodes and esx hosts.  And yes my san providor does recommend disabling spanning tree for the entire vlan, but I did not.

That said, yes the 2x 10G are in a port-channel, there are no logs indicating any issues with the port channel on the other switch, but there was another switch adjacent to the one that had issues and it showed the interface going up and down, so I assume that all of the interfaces were going up and down. 

Still wondering if there is any other saved information on the switch after a reboot.

Also I guess the port LED's on these 4948's dont blink?

Dan.