Link Active WIth No Preceding Loss of Link

[Amtiskaw]

Hi

I have a 8212zl  running K.15.06.0008 that I'm seeing some odd behavior on. We run a stack of 5 VMWare hosts, each having 5 NICs connected to this 8212zl. The VMWare management client is occasionally reporting loss of connectivity on several NICs on several hosts at once.

In Procurve Manager, I see a "Link active on port xx" message when the VMWare says a particular NIC has come back, but there's no corresponding "Loss of link on port xx" preceding this that matches when VMWare said the NIC went down.

The outages are typically less than a minute.

Has anyone come across this sort of behavior before?

Cheers

Simon

[Amtiskaw · Accepted solution]

For the record, this was happening because a J9534A 24 port copper module was faulting. Debug logging showed the chassis losing heartbeat from the module. Then the ports would go down with no warning, and come back up when the module started behaving again.

This started happening after upgrading from K,15.04.0003 to K.15.07.0008. I've rolled back to the secondary image with the older firmware until we get a replacement module, then I'll try K.15.07.0008 again.