Loss of stack member

[Tony Barrett_2]

I have quite a few stacks (4-6 switches in each) of ProCurve 2650's. The stacks are connected via the copper Gb ports on each switch, with the commander having the Gb fibre module and feed back to the core switch (5308).

The commander does not auto grab, and the members do not auto join. All are configured manually, and show successfully joined to the stack. All switches are running the latest f/w.

What I am seeing fairly regularly on some stacks (but not all) is the error 'Loss of stack member on port: xx' where port is either 49 or 50 (the stack uplink port). The errors are being logged to the switch log and to OV NNM via SNMP.

When I check the switch stack, all shows ok. There are no faults observed on any of the switches. Some users are reporting that when this happens, there is momentary loss of connection (enough to drop an SSH connection), then things resume as normal.

I can't seem to find anything that may indicate a hardware fault, and the config on the switches is pretty standard.

Anyone have any ideas?

[Gonzo Granello]

Tony,

check the switch event log and verify that the ports involved are up all the time. Setting the time on all switches will help to verify when and what happens. I have a 4108 here that experiences random port toggeling - the log shows port a2 down /up as example for no apperant reason and i know for sure that the connected devices are up all time. It seems that the latest firmware also has no effect as to fix the issue and as far as I know HP had those issues for quite some time with a lot of their switches. There are unfurtunatley not many options until the issue get fixed, but i guess a call to support wont hurt either.

GG