4.0.0U1: Can't (re-)init Storage LUN after crash

[fasteddy]

Hello,

after a crash the LUN1 of my Infortrend-Storage crashed, so I had to reconfigure it. I vMotion'ed the VMs to another place and correctly unpresented all LUNs from 0 up to 2 (esxcli corestorage commands...). Then I re-partitioned the Disk (to be sure I took different sizes) and remapped the 3 prtitions again to LUNs 0, 1 and 2.

I first initialized LUN0 as "Shared storage 1" and wanted to init LUN1 as "Shared storage 2". But now I am sitting here waiting for the LUN to be loaded (see image, meanwhile the ESX found the partition but opened dialog, that tells me: <see error.png>).Trying to format manually with

vmkfstools -C vmfs3 /vmfs/devices/disks/naa.600d0230ffffffff00000230ede95101

results in the error message: Failed to check for existing file system on device '/vmfs/devices/disks/naa.600d0230ffffffff00000230ede95101'.

Is there another possibility to solve this problem? I thaught repartitioning with different sizes than before would have been enough to re-layout the disks, but it does not seem so. Maybe this could be caused by a defective disk? But the storage does not show up any error.

Best regards, Stefan

[binoche]

1, please post the result of fdisk -lu /vmfs/devices/disks/naa.600d0230ffffffff00000230ede95101

2, vmkfstools -C vmfs3 /vmfs/devices/disks/naa.600d0230ffffffff00000230ede95101 is not correct, you need to partition it first and then such as vmkfstools -C vmfs3 /vmfs/devices/disks/naa.600d0230ffffffff00000230ede95101:1

binoche, VMware VCP, Cisco CCNA

[fasteddy]

Sorry, but I am a little late on reading. Because the system is a production environment, I had to find a fast way to solve this problem...

Indeed 'fdisk -lu' reported al but not the problematic LUN, or in other words: the normally mapped linux device /dev/sd... was not listed as a result, hence there was nothing to partition. And what way ever I used to un- and represent the LUN,, it kept unmapped. So last and least I resized tha partition up to the combined size of the former two and represented that as a completely new LUN... and there it appeared fast and functionable. It just had to be reformatted after vMotion'ing the still intact VMs. Now and in future, when working on an Infortrend device, I will not use again Storage-side partitioning, but I will create and present all LUNs as unpartitioned logical drives, no matter what storage space loss is caused by the RAID data.

Nevertheless, thanks for your answer.

With best regards, Stefan.