Redundant ISCSI Topolgy problem with 8212 switch

[ahmeteris]

We have topogly which at the attached.

- We have 2 vlan which is one ISCSI and the other is Local Network.

- Hp blade server have two Flex-10 Modul with one ports connect 10G to the switch untagged port for ISCSI, the other ports connect 10G to the switch untagged port for LAN. Two 8212 connect with 10G(it didn't see at topology picture)

- All Storage node connect to twice 8212 switch with 2 10G cable. One cable connect to one 8212 and the other connect to other 8212. ALB (adaptive load balancing) config has been made at all storage node.

- Jumbo frame support enabled switchs at ISCSI vlan.

- Two 8212 connect to uplink switch Hp 2910 with 10G connection. Spanning tree enable 3 switchs.

- There aren't any additional config at the switchs.

We have some problem, the system work good some times but after more the system will be down. some pings broken and system will slowly. I think may be there are a loop in system. But ı am network eng. so dont understand storage and server.

Are there any advice for this system? Are there aynoen which have any experience with ISCSI topolgy? Pls help me?

[RemiBatist]

Hello,

Is the ALB an active/passive config?

Can you tell which the spanning-tree root switch is?

did you enable jumbo frames also on the flex-10/os/hypervisors and the p4500?

Is flowcontroll enabled on all the connected ISCSI ports on the switch and flex-10?

Remi

[Jay Cardin]

It sounds like Spanning Tree Protocol is blocking the 10G connection between the two 8212's.  If that were to happen, you would essentially have created two seperate ISCSI networks and with ALB enabled, you would probably see slowdowns and loose pings.

On each 8212, connect to the console and run: show spanning-tree

Look for the interface that connects to the other 8212.  Does either side show Blocking?