Background:
Host is an R730 running 6.5u2 Build 8935087 with a ~3TB VMFS-5 datastore
Recently added 4x1TB SATA drives to the RAID5 array via OMSA, reconstructed, and am ready to increase the Datastore size.
The problem is that this is what I see in the GUI:
This just makes zero sense to me. I've gone through the logs and didn't see any errors that point to a problem. I've verified the GPT is not corrupt and I'm just at a loss here. I could go through and grow it through tthe CLI, but since this IS a prod box, I'm not too keen on rolling the dice.
Any ideas?
That's what I was afraid of. The partition table is not GPT (required for partitions >2TB), so your options are limited. The automatic MBR-to-GPT conversion may not work due to the second (Dell diagnostic) partition, i.e. it needs to be done manually.
The most secure - and likely only supported - way to do this, is to backup everything, reformat the whole disk, and then restore the VMs again. It may however also be possible to overwrite the current partition table with the proper options (i.e. create a GPT partition table, and preserve the VMFS partition) to avoid the restore (the backup should be done anyway!!). I'm not 100% sure whether this will work, so be prepared for the required restore time.
André
---------------------------------------------------------------------------------------------------------
Was it helpful? Let us know by completing this short survey here.
That diagram suggests that it's VMFS6, not VMFS5. How does it show in the vSphere Client (the HTML5 interface, rather than Flex)?
That's what confuses me.
That screenshot is taken from the Expand Wizard in the HTML5 web client.
Here's some more info:
[root@***-******:~] esxcli storage filesystem list
Mount Point Volume Name UUID Mounted Type Size Free
------------------------------------------------- ----------- ----------------------------------- ------- ------ ------------- ------------
/vmfs/volumes/57bb669d-2c7c9511-f87e-246e960b7238 Data 57bb669d-2c7c9511-f87e-246e960b7238 true VMFS-5 2198754820096 372488798208
ok, conversely how does it look in Flex client?
Do you mean the old vSphere Client? If so, the host is on 6.5u2 so I can't connect to it with the client anymore.
Don't trust the GUI, it always shows VMFS6 for existing datastores regardless of its current format.
It may indeed be the safest (and likely only) way to increase the VMFS partition from the CLI in this case.
Please run partedUtil getptbl /vmfs/devices/disks/<device-name> for the LUN to see how it really looks like. I assume that the first partition is a Dell diagnostic partition.
André
PS: You mentioned a ~3TB VMFS datastore. However to me it looks like this is a 2TB datastore due to a MBR partition table.
Here's the output:
[root@***-*******:~] partedUtil getptbl /vmfs/devices/disks/naa.614187706cdb33001edff5fd1b82bc23
msdos
850740 255 63 13667139584
1 63 80324 222 0
2 81920 4294961684 251 0
Partition 1 is indeed a Dell Diag Partition
That's what I was afraid of. The partition table is not GPT (required for partitions >2TB), so your options are limited. The automatic MBR-to-GPT conversion may not work due to the second (Dell diagnostic) partition, i.e. it needs to be done manually.
The most secure - and likely only supported - way to do this, is to backup everything, reformat the whole disk, and then restore the VMs again. It may however also be possible to overwrite the current partition table with the proper options (i.e. create a GPT partition table, and preserve the VMFS partition) to avoid the restore (the backup should be done anyway!!). I'm not 100% sure whether this will work, so be prepared for the required restore time.
André
---------------------------------------------------------------------------------------------------------
Was it helpful? Let us know by completing this short survey here.
Whoah... ok... that's weird.
I would have SWORN that thing showed GPT and not MBR yesterday...
Damn. Yeah, that'd be the issue.
What about this:
1) Create a new datastore in the 4TB of freespace that's gpt and VMFS6
2) Migrate the VMs to the new datastore
3) Kill the old datastore
4) Extend the VMFS datastore to encompass all the space
Is that possible or does the whole thing need to be nuked?
Thanks all for the help!
Wait, scratch that.
It's the entire partition table that needs wiping and rebuilding.
Damn.
Oh well! Looks like a trip to the remote site is in the cards soon! Thanks for the help!
I'm finally getting around to testing some solutions to this problem without a ton of downtime. I've created a test ESXi 6.5 box and mapped out the same partitions on the disk. I used this command: partedUtil setptbl "/vmfs/devices/disks/naa.6a4badb00d3ad900251513900ec62218" msdos "1 1 69631 222 0" "2 69632 682053631 251 0" Now... the GUI is showing some jacked up numbers EVEN after a reboot. What the hell is going on here?
Where did you find these partition start, and end sectors? They don't match with those reported earlier.
André