Insufficient disk space on datastore

[admin]

hi guys,

weird problem... we've got a VM which has RDM's attached. Now i powered off the machine and i would like to relocate the C-drive (just a normal vmdk) to another datastore.

the vmdk itself is about 11G and free space on datastore i want to copy to is around 50G. Although i'm getting the error "Insufficient disk space on datastore"

There are 4 rdm's attached (physical compatibility) 2 x300G and 2x40G

Anybody a clue what this might be?

The VM:

\[root@tonic knife_vm_play001]# du * -sH

11GB bobolink

Datastore i want to copy to

/vmfs/volumes/456ac485-fb669a70-5ece-001143e11b57

99G 46G 53G 46% /vmfs/volumes/washer_vm_play005

[xAyiDe]

What filesystem is the datastore?

I mean are you trying to copy a vmdisk from one vmfs to another. Or from an vmfs to some ext2/3?

null

[sbeaver]

I think ESX is tricked and think the RDM's are vmdk with those sizes and will not fit on the datastore you trying to copy to. I would use vmkfstools to move the vmdk and update the vmx manually

[admin]

it's just a normal vmfs3 volume..

and to sbeaver.. i'm also thinking that this might be the problem. Last night we set up an extra lun of about 400G, we're trying to migrate to this lun.. if this works without a problem, i think we found some kind of bug.. and will create a SR

i'll try to post the details within a few hours

[admin]

seems that on our 400G lun it starts copying without a problem.. it's rather slow so i guess that he is moving the 300G RDM??? strange to me because it's an rdm and nog a vmdk..

will open an SR with vmware for this one

[VirtualKenneth]

Can't you detach the RDM('s), move the VMDK and reconnect the RDM's?

[admin]

offcource this can be done... but for me this is finding a way around the problem and to get a quick solution.. it's not a solution for the problem itself..

[VirtualKenneth]

Correct, it's only a workaround though.. please let us know what VMware has to say about this.

[VirtualKenneth]

Confirmed, I've tested to clone a RDM vmdk to a LUN that is smaller then the size LL gives back. This doesn't work.

This must be a bug since cloning a RDM vmdk of 5 GB goes way faster then a real vmdk of 5 GB so there is some intelligence anywhere

I knew there was a command to see the exact absolute disk size of the RDM file, anyone who remembers this?

[admin]

Currently we've submitted a SR for this issue and vmware support is consulting the Research team to solve the problem...

hopefully it's solved very soon

[jkramp]

hi Tom,

did you receive any response on youre SR?

I have the same problem and hope to find a proper solution.

[ErikErik]

they are still investigating... it seems they need a lot of time to solve this one... haven't got any response from tech support...

Tom

[admin]

ok... final answer from vmware support.. there will be a new KB about this problem. It should be released shortly.

solution for now.. until a new solution arrives..

\- delete the rdm from the powered off vm

\- migrate the VM

\- attach the rdm again