Just updated an existing ESX 3.0.1 server to add 2x HP FCA2214, aka QLA2340 single channel, 2 Gb cards.
I'm already running 3 other servers against 4 luns on an EVA 8000.
After the san admin zoned the system in, everything was fine. 4 luns, 4 paths on the SAN, 2 HBA, so 8 paths per lun, 4 on each hba.
Then the fun started. After flipping back and forth among servers comparing configurations, I noticed there were no luns showing up on vmhba1. I refreshed, then rescanned a few times. The luns showed back up, but with only 3 paths to the san on vmhba1. There are still 4 paths on vmhba2.
Any ideas? Here are the outputs of esxcfg-mpath -l for this system and another system to compare.
Broken (7 paths):
Disk vmhba2:0:1 /dev/sda (409600MB) has 7 paths and policy of Fixed
FC 10:1.0 210000e08b8e9597<->50001fe15006b238 vmhba2:0:1 On active preferred
FC 10:1.0 210000e08b8e9597<->50001fe15006b23a vmhba2:1:1 On
FC 10:1.0 210000e08b8e9597<->50001fe15006b23c vmhba2:2:1 On
FC 10:1.0 210000e08b8e9597<->50001fe15006b23e vmhba2:3:1 On
FC 6:1.0 210000e08b8ec197<->50001fe15006b23b vmhba1:0:1 On
FC 6:1.0 210000e08b8ec197<->50001fe15006b23d vmhba1:1:1 On
FC 6:1.0 210000e08b8ec197<->50001fe15006b23f vmhba1:2:1 On
OK (8 paths) this system has single dual-port card
Disk vmhba2:0:3 /dev/sde (30720MB) has 8 paths and policy of Fixed
FC 2:1.0 210000e08b8b9eaa<->50001fe15006b238 vmhba2:0:3 On active preferred
FC 2:1.0 210000e08b8b9eaa<->50001fe15006b23a vmhba2:1:3 On
FC 2:1.0 210000e08b8b9eaa<->50001fe15006b23c vmhba2:2:3 On
FC 2:1.0 210000e08b8b9eaa<->50001fe15006b23e vmhba2:3:3 On
FC 2:1.1 210100e08bab9eaa<->50001fe15006b239 vmhba1:0:3 On
FC 2:1.1 210100e08bab9eaa<->50001fe15006b23b vmhba1:1:3 On
FC 2:1.1 210100e08bab9eaa<->50001fe15006b23d vmhba1:2:3 On
FC 2:1.1 210100e08bab9eaa<->50001fe15006b23f vmhba1:3:3 On
Any ideas?
Also, on this system, doing a rescan from the command line yield different results on different nics:
\[root@tekvm1 log]# esxcfg-rescan vmhba1
Rescanning vmhba1...done.
On scsi0, removing:.
On scsi0, adding:.
\[root@tekvm1 log]# esxcfg-rescan vmhba2
Rescanning vmhba2...done.
On scsi1, removing: 0:0 0:1 0:2 0:3 0:4.
On scsi1, adding: 0:0 0:1 0:2 0:3 0:4.
\[root@tekvm1 log]#
Swapped LUN IDs on the problem system seemed to contribute.
Hi, I occur the same problem with ESX 3.0.1 ... Is it a bug ?
"felt" like some kind of inconsistency in the drivers, but can't know for sure. All I can say is today there are 8 paths to each lun.
I will also say that with the 2340 and the HP server the luns will not show all info in the storage section under configuration. The 2462 in the ibms always shows the number of paths and status. The hps don't always. I have to rescan to get that info.
To me that indicates something is conflicting in the 2340 with HP servers. But both HP and VMware closed the case when we couldn't find anything wrong in the configuration other than one path would come and go.