VMFS volumes not seen MSA 1500

[migeauxy]

Hi Folks

We have a configuration like this one :

2 x DL380 G5 Servers (ESXA and ESXB) with 2 Qlogic Single port HBAs (vmhba2 & vmhba3) running ESX 3.0.2

these servers are connected through HP FC switches (SWA & SWB) to an MSA 1500CS disk array, with Active/passive controllers and 2 FC connections at the back of the box

Cabling is as follows :

ESXA-vmhba2 -- SWB-port3

ESXA-vmhba3 -- SWA-port2

ESXB-vmhba2 -- SWA-port3

ESXB-vmhba3 -- SWB-port2

SWA-port1 -- MSA1500-FCconnector"left"

SWB-port1 -- MSA1500-FCConnector"right"

For the moment, path policy is set to fixed for 3 existing luns.

When I try to create another LUN, I can see it in the Storage as a LUN, but I cannot format is witha VMFS format.

What I did is change configuration of this LUN so that its path is on the same as the other ones, then I can format.

What I can't explain is why this LUN has taken by default the "passive"controller of the MSA as it's active path.

Anyone has an explaination ?

Thanks for your support

here is the mpath -l

# esxcfg-mpath -l

Disk vmhba0:0:0 /dev/cciss/c0d0 (34699MB) has 1 paths and policy of Fixed

Local 6:0.0 vmhba0:0:0 On active preferred

RAID Controller (SCSI-3) vmhba2:0:0 (0MB) has 1 paths and policy of Most Recently Used

FC 11:0.0 50060b000065e53c<->500508b300916bc9 vmhba2:0:0 On active preferred

Disk vmhba2:0:1 /dev/sda (92118MB) has 2 paths and policy of Fixed

FC 11:0.0 50060b000065e53c<->500508b300916bc9 vmhba2:0:1 On

FC 23:0.0 50060b000065e66c<->500508b300916bc1 vmhba3:0:1 On active preferred

Disk vmhba2:0:2 /dev/sdb (233966MB) has 2 paths and policy of Fixed

FC 11:0.0 50060b000065e53c<->500508b300916bc9 vmhba2:0:2 On

FC 23:0.0 50060b000065e66c<->500508b300916bc1 vmhba3:0:2 On active preferred

Disk vmhba2:0:3 /dev/sdc (233966MB) has 2 paths and policy of Fixed

FC 11:0.0 50060b000065e53c<->500508b300916bc9 vmhba2:0:3 On

FC 23:0.0 50060b000065e66c<->500508b300916bc1 vmhba3:0:3 On active preferred

Disk vmhba2:0:10 (140011MB) has 2 paths and policy of Most Recently Used

FC 11:0.0 50060b000065e53c<->500508b300916bc9 vmhba2:0:10 On active

FC 23:0.0 50060b000065e66c<->500508b300916bc1 vmhba3:0:10 On preferred

[java_cat33]

My understanding is that if you have active/passive controllers - ESX should detect this and your paths should be set to MRU. Have these settings been changed?

[Erik_Zandboer]

Hi,

As far as I know the MSA1500 has two controllers, and each controller has an internal FC connection to the switch mounted directly opposite to it (in the back). So only the active controller would use a single FC switch in the back, and therefore all paths should run through that controller and FC switch. MRU settings should be applied at the ESX level.

I heard that msa1500 now supports active/active as well, but as ESX does not, it will not help you (a single LUN can only be seen through the one and same controller, and as storage in ESX terms is usually R/W to all ESX hosts this limitation applies)

[java_cat33 · Accepted solution]

Yes it does support active/active now with their latest firmware. ESX also detects it and sets the path to fixed - works with no problems

[migeauxy]

Hi Guys

Following your recommendations, I flashed the MSA firmware to Active/active . works now like a charm

Tried the MRU policy with Active/Passive configuration, and it seems not to work in all "problem" cases. for example : if a FC cable is unplugged at the FC switch level or at the MSA Controller Level, the MSA controller thinks he is OK and doesn't switch to anothe path.

with Active/active, it works, so we let it working this way

Thanks for your precious help

[Schorschi]

Double check the HP docs? EVAs HP often recommends Fixed. I have active passive on MSA1000 in our crash and burn lab, active/passive running 5.20 just fine. We have older MSA1000s running 4.48 in other labs using MRU just fine as well. Is your zoning setup identical across both FC switches?