Hello to all.
Now have some strange problem with spanning-tree on sps224g4 switches (organization owns 50 such devices, all identical one with another, with the same hardware and software versions, and all of them were bought at one time).
Without any outer factors (without links up/down, operator logins, etc..) spanning-tree on this devices may be TURNED OFF by switch itself.
At the same time, BPDU control becomes to blocking state (bpdu blocked on ports, where stp turned off), e.g. all bpdu's dropped at device due to global stp disable.
There is no need to tell, what happens after that, network just become layer2 looped.
No telnet/ssh sessions can be established at this time, because of 100% bandwidth utilization (1Gbps traffic in looped network can be reached due to only one "arp who has" request). In result at least 10 access switches in one layer2 segment become unreachable in way of peak load of physical interfaces.
After that, the only solution is to disconnect switch from network by removing cables from its ports manually(to break the loop physically). When situation normalizes in this (physical) way - no leading messages can be located in switche's logfile.
In term of 10 monthes this happen 5 times with 5 different switches in 4 different layer2 segments.
Maybe, somebody have the same problem to?
Regional Cisco representatives can't comment this, except constanting a fact, that new firmware available. But, is this problem known? Is it fixed? I don't want to do some upgrades without necessity and 100% ensurance. And, as far as I tested manually, at least one problem (with combined policy-maps) wasn't resolved in latest firmware. So just hope for normal work is not a reliable solution.
10.255.0.41# sh version
SW version 1.0.1 ( date 29-Nov-2007 time 15:58:42 )
Boot version 1.0.2 ( date 13-Nov-2007 time 14:11:51 )
HW version 00.03.00