HP Procurve 9315 + 2650 Network down RSTP or VRRP ?

[Thijs_1]

We are currenlty building a new HP Procurve network. Two 9315 switches in the core, trunked together. A lot of 2650 edge switches connected with a fiber uplink to both core switches (on port 49 & 50). We are using RSTP to prevent a loop. One of the coreswitches is RSTP root bridge and has a RSTP priority of 4096. The other core switch has a priority of 8192 (backup RSTP master). The edge switches have their default RSTP priority. We are also using VRRP and about 25 VLANs. The switch that is RSTP root is also VRRP master for all the VLANs. The other core switch is VRRP. backup. There is a 1 GIG connection between our HP core switch and our old 3com corebuilder 9000 switch. This is a Layer3 routed connection.

The entire migration was going great, until about 2 weeks ago. It seems like, when we migrate another site, and connect the HP switch 2650 to the HP 9315 (only one fiber uplink), the RSTP topology change that is generated blocks all PC traffic on switches in a certain VLAN (not all the switches in that VLAN by the way.). So far HP and our local supplier can not find the issue.

The software version on our 2600's : H7.50
The software version on our 9315's : 07.8.00

A sample of an edge switch config is attached.
Any help will be appreciated.

[Arnaud_9]

Hi,
Here are some ideas, I hope one will help you :
Did you try to connect this new switch with all users port disabled (only the uplink enabled) ?
If it works, reenable all ports, and check if the root switch is the same before and after : A switch you don't know could be connected to the new 2650, somewhere in the building, with a lower priority. This switch can become the root and maybe could change all the topology.
Then, your connectivity problem sounds like an inconsistant VLAN propagation on blocked links. It could explain that some users on a VLAN are working on some switches, and not on other switches : Everything work fine when the main links are forwarding, but when there is a topology change and blocked links become forwarding, the traffic on a vlan cannot be forwarded to some switches.

You can check the root switch with "show 802-1w" on 9315, and a "show spanning-tree" on 2650.

Then, you could also check if the VRRP master is always the only one owner, by checking the "state" field of the "show ip vrrp" command on both 9315.
The trunk between 9315 should never be blocked, and should have all VLANs whith VRRP enabled.

[Thijs_1]

At the moment we have and engineer of our supplier and a HP engineer at our site. We have been troubleshooting all day (16 hours) and it seems to have something to do with the 4 Gig trunk between the core switches.

When slot 1/1 to 1/2 on active supervisor Core switch A and slot 2/1 to 2/2 on backup supervisor Core switch A is connected to the corresponding supervisors in core switch B, the are problems on this one VLAN (we us single instance spanning tree) If we remove the trunk on the two backup supervisors in both core switches everything seems to work fine .....


Still no solution.


Will be continued.