ESXi 4.1 on HP Blade (BL460c G7) in BladeSystem c7000 Enclosure G2 with HP VC FlexFabric 10Gb/24-Port Module and Netapp FAS 6240 Dead Paths on 10GB FlexFabric CNA for one of the Netapp Clustered Heads.

[shepnasty]

ESXi 4.1 on HP Blade (BL460c G7) in BladeSystem c7000 Enclosure G2 with HP VC FlexFabric 10Gb/24-Port Module and Netapp FAS 6240 Dead Paths on 10GB FlexFabric CNA for one of the Netapp Clustered Heads.

I can see the paths to one of my Netapp heads but the other head for failover shows dead paths.  After troubleshooting the issue it is isolated to VMWare ESXi 4.1 Blade Hosts in the HP enclosure.  Windows Blades in the same enclosure see the paths back fine using Netapp DSM.  I am also unable to add a lun from the head where the paths are dead (obviously).  I have provissioned storage from that head everywhere else in our environment.

Any ideas?  Is this an ESXi thing?  Our main VMWare cluster is IBM ESX 4.0 hosts and they see the storage just fine.

SHEP

[shepnasty]

We were able to replicate the issue and resolve it.  It seemed to be an issue with ALUA.  We disabled ALUA on the Netapp Igroups and then enabled them again and all seems to be fine.  We are using RR path selection.

SHEP

[tjaster]

We are also facing this problem. But nothing helped. Our NetAPP Luns are all running with ALUA and RR. It seams to be a software bug in the FC/FCOE driver and  be2net driver in combination. Did you still drive this configuration? What software versions are you using. See ours attached.

[mitchellm3]

Did you ever get this resolved?  We are working with HP, VMware and IBM (storage) trying to figure this out.  All signs point to the onboard CNA or FlexFabric but we are not sure.  However, our LUNs seem to lock...we don't necessarily lose paths as far as we can tell.  After last nights outage, we have enabled verbose logging on the Emulex adapter as well as the ESXi host.

[Shocko]

I have seen the same issue on one of my enclosures runing ESX 4 U3. I'm sure if I logged it to HP they would tell me to update the enclosure frimware though! When I did a rescan though, it didn't pick up the dead paths at all i.e. it didn't pick up anything on the HBA with the issue. Since these are G6 blsys460 blades, the HBA is a single mezz card so it would be unusual for a single port on it to fail (if this is even possible) . Also, all physical links are showing as up so it seems like a firmware/software issue. I'm going to reboot the ESX host and see what happens.