[c-nsp] 65th VLAN

[Vincent De Keyzer]

Hello fellow list members,

 

we had a migration on our main Ethernet network last week that evolved into
a real disaster.

 

A whole VTP domain had gone bananas - we eventually had to power down most
of the switches, and then power them back up one by one in order to recover
stability.

 

Of course, it's hard to find out what caused the problem now that all
suspects have been rebooted - but there is one thing I would like your
opinion on.

 

The day before the migration, a human mistake had led to the configuration
of a 65th VLAN on a switch (2950T-24) that supports only 64 STP instances
(although it can support 256 VLANs). Nothing happened then, so nobody
noticed; but when the migration started the day after, as cables were
disconnected and reconnected between switches, Hell was unleashed. 

 

Some specifics: 

*      At some point, we saw some things working (read: pinging) with a
regular pattern: only 10 seconds every 30 seconds

*      A 3524XL (which has the same limitations on VLANs) also had 65 VLANs
at some point because it learned the 65th VLAN via VTP.

 

Did anyone ever experience a very bad outage due to more VLANs than
available STP instances? What is the behaviour of a switch (2950/3524) when
it has exhausted its STP instances ? Does the pattern of 10+20 seconds ring
a bell to somebody?

 

Any hint will be appreciated.

 

Vincent