Multipathing problem

[rierajf]

Hi,

I recently installed a new ESX 3.0.2. It is connected to a HDS 9585 array through fabric SAN. But, the problem is the host does not identify multipaths to LUNs as multiple paths but it shows them as multiple LUNs. So, there is no redundancy for the SAN connectivity.

For example, the result of command vmkpcidivy -q vmhba_devs :

vmhba0:0:0 /dev/cciss/c0d0

vmhba1:0:100 /dev/sda

vmhba1:0:101 /dev/sdb

vmhba1:1:102 /dev/sdc

vmhba2:1:76 /dev/sdd

vmhba2:1:99 /dev/sde

vmhba3:0:102 /dev/sdf

vmhba3:1:100 /dev/sdg

vmhba3:1:101 /dev/sdh

vmhba4:0:76 /dev/sdi

vmhba4:0:99 /dev/sdj

Has someone experienced something like that ? I can not go in production without redundancy to SAN connectivity.

Thanks a lot,

JF

[dominic7]

What output do you get for: esxcfg-mpath --list

?

[sbeaver]

Chances are zoning and masking are not exactly the same on the SAN for each of the ESX servers and ESX then gets confiused. Douple check your SAN configuration.

[rierajf]

Here the result of command esxcfg-mpath --list :

Disk vmhba0:0:0 /dev/cciss/c0d0 (69973MB) has 1 paths and policy of Fixed

Local 8:0.0 vmhba0:0:0 On active preferred

Disk vmhba1:0:100 /dev/sda (102400MB) has 1 paths and policy of Most Recently Used

FC 66:1.0 10000000c9621f4a<->50060e8000c3df12 vmhba1:0:100 On active preferred

Disk vmhba1:0:101 /dev/sdb (102400MB) has 1 paths and policy of Most Recently Used

FC 66:1.0 10000000c9621f4a<->50060e8000c3df12 vmhba1:0:101 On active preferred

Disk vmhba1:1:102 /dev/sdc (102400MB) has 1 paths and policy of Most Recently Used

FC 66:1.0 10000000c9621f4a<->50060e8000c3df16 vmhba1:1:102 On active preferred

Disk vmhba2:1:76 /dev/sdd (40960MB) has 1 paths and policy of Most Recently Used

FC 66:1.1 10000000c9621f4b<->50060e800043d2a3 vmhba2:1:76 On active preferred

Disk vmhba2:1:99 /dev/sde (51200MB) has 1 paths and policy of Most Recently Used

FC 66:1.1 10000000c9621f4b<->50060e800043d2a3 vmhba2:1:99 On active preferred

Disk vmhba3:0:102 /dev/sdf (102400MB) has 1 paths and policy of Most Recently Used

FC 66:2.0 10000000c9621f48<->50060e8000c3df13 vmhba3:0:102 On active preferred

Disk vmhba3:1:100 /dev/sdg (102400MB) has 1 paths and policy of Most Recently Used

FC 66:2.0 10000000c9621f48<->50060e8000c3df17 vmhba3:1:100 On active preferred

Disk vmhba3:1:101 /dev/sdh (102400MB) has 1 paths and policy of Most Recently Used

FC 66:2.0 10000000c9621f48<->50060e8000c3df17 vmhba3:1:101 On active preferred

Disk vmhba4:0:76 /dev/sdi (40960MB) has 1 paths and policy of Most Recently Used

FC 66:2.1 10000000c9621f49<->50060e800043d2a0 vmhba4:0:76 On active preferred

Disk vmhba4:0:99 /dev/sdj (51200MB) has 1 paths and policy of Most Recently Used

FC 66:2.1 10000000c9621f49<->50060e800043d2a0 vmhba4:0:99 On active preferred

[rierajf]

Thanks for your responses.

I have only one ESX at the moment.

I have noticed that each time I reboot the host, LUNs are viewed like snapshots. Then I forced LVM.DisallowSnapshotLun to 0 in order to view my datastores.

I have HP DL585 G2 with Emulex HBA, brocade SAN switchs and Hitachi HDS 9585 array for information.

JF

[sbeaver]

It comes down to zoning and masking on the SAN each path is not presented the same and ESX sees that as 2 different LUNS.

[rierajf]

Thanks everybody for responses.

On HDS9 arrays, we must put the Sun Cluster Connection Mode parameter for the ESX host groups.

It is OK with this parameter now.

Sincerely,

JF