Unable to format a Lun with VMFS

[opbz]

HI

I am having a strange problem and was wonderring if anyone else has seen it or has any suggestions.

I have several servers attached to a CX300 with flare 26.

All servers are running ESX 3.5 with ALL patches. There is also a VC server running 2.5

At moment we have all only one server running. It is connected to the CX using 2 qlogic cards. THe CX is configured to give failover mode 1 even though its flare 26 and could potentially use failover mode 4. But I know Vmware do not recommend failover mode4 yet.

Problem we have is we are able to see the luns but when we try to create a VMFS datastore it nearly always fails. We created 2 datstores of about 15. THe datastores are all less than 500GB.

If we go into console we can see it creates a vmfs partition on the luns but it does not format or give it a valid label it never appears as a datastore and is not listed under /vmfs/volumes. To be able to use the lun again I have had to manually edit the file table and delete the partition created.

The CX appears to be fine we do not see any errors on itr or on the connected switches.

All the servers are fresh installs, no upgrades all installs are done using tested media (iso images checked by md5sums) all hardware is on supported hw lists.

any suggestions? any ideas?

[java_cat33]

I've had a similar issue on a HP MSA 1500 SAN. I got around it by deleting the lun, re-creating it and re-presenting it.

May work for you?

[mike_laspina]

Hello,

Sometimes it is necessary to zero the first 64KB of the partition before it will format with vmfs.

run esxcfg-vmhbadevs to see which device the vmfs volume is on.

or use fdisk -l an look for the fb id.

then zero it with

dd if=/dev/zero of=/dev/sdX bs=512 count=128

[opbz]

thanks guys...

we tried both options with unfortunatelly no luck. We are able to create partitions jut not formatted VMFS ones..

We even presented the luns to other servers on the san and we where able to format them both in linux boxes and windows boxes...put them back on the ESX servers and we are again stuck. WE cadelete whats there but thats about it

We got vmware involved and they said it was hardware san related. Not sure where they got that idea from taking into account that the luns worked perfectly when attached to non vmware servers.

might be a problem with flare 26 which is supposed to enable CX's SPs to be presented as active active something still not supported by vmware

[jeffjprince]

I recently ran into this exact problem with a little bit different hardware.

2 X Dell R710 servers

2 X Brocade switches

Dell Branded EMC CX4-120

Error everytime I try to format luns presented from the CX4. I started by hopping on the EMC chat support, and got no where. They blamed VMware. Stating that if you can see the lun, then the Clariion did its job.

So I called in a ticket to VMware. The tech there worked with me for a couple hours trying everything from reboots, to recreating the lun, to checking the zoning. Nothing worked. Still got the same error when you try to add the disk from virtual center. He asked that I call EMC back and setup a conference call between them both and I.

I called EMC, and worked with the rep there for a couple hours. He pulled my logs, checked settings etc. Got no where.

So I setup a conference call to have them both work together to find an answer.

• THE FIX ***

Before the call happened, I tried upgrading the FLARE from version 6.28 - 6.29. Boom fixed. I am really suprised the EMC rep didn't suggest this.

Jeff