STP in IBM NOS, how?

Question

Computer networks

Greetings.
There is a situation like this: An
IBM Blade Center H blade chassis, it has two IBM BNT Virtual Fabric Switch Module switches with identical settings, let's call them BNT1 and BNT2 for clarity. Each of them is connected by one link to the Cisco 6509. To the heap, there is a direct link between these two BNTs. To disable redundant links and avoid loops in the network, the switches use PVRST, which has 128 Spanning Tree Groups. As new Wealans are added to the switch, the number of free STGs has dried up, and all new ones are automatically added to the default STG number 1. Problems began when the seventh Wealan (in this group) got into stg 1 on BNT1. Symptoms of the problems are as follows:
1. The assignment of ports in the group changes every second. It should be like this:

EXT1           128    20000!   DISC  ALTN мак-адрес соседнего коммутатора     8011  P2P - линк к соседу

EXT5           128    20000!   FWD   ROOT мак-адрес Cisco 6509     9116  P2P - линк к шеститоннику

And it happens like this:

EXT1           128    20000!   FWD   ROOT мак-адрес соседнего коммутатора     8011  P2P

EXT5           128    20000!   DISC  DESG свой собственный мак-адрес     8015  P2P

Or like this:

EXT1           128    20000!   DISC  DESG мак-адрес соседнего коммутатора     8011  P2P

EXT5           128    20000!   FWD   ROOT мак-адрес Cisco 6509     8916  P2P

Or even like this:

EXT1           128    20000!   DISC  ALTN мак-адрес соседнего коммутатора     8011  P2P

EXT5           128    20000!   DISC  DESG свой собственный мак-адрес     8015  P2P

2. A six-ton ​​man sees his own frames on both interfaces with which he is connected to these BNTs.
3. BPDUs from BNT1 come to the six-ton ​​on both ports.
4. In the wealan, which is the seventh in a row, regardless of its VLANID, packets go through one. One answer - one lost.
Does the respected community have any ideas how to deal with this without rebooting the switch? And then I ran out of ideas, and only the thought remained that this was a glitch that would be cured by a reboot.

Answer 1

128 STP instances is a lot, a lot. Including for the six-ton. If there is no reason why you need to keep the trees different for each VLAN, switch to MST, since judging by the docks, NOS supports it. Then there will be a pair of STGs for at least a thousand VLANs, with instant convergence.

all new ones are automatically added to the default STG number 1

This is a bad situation, unpredictable. MST is designed to keep one tree per bunch of VLANs even when docking with PVRST (but the root must be within the MST area). In your case, this is not the case. Any TCN can blow away the roof of an IBM piece of iron. This seems to be happening. I strongly doubt that a reboot will solve the problem - your configuration is unsupported.