Impossible to re-Attached an extended vRDM disk after detaching it.

[hpoirier8]

Hi,
Our VMware version is vSphere 6.0 MU3.
We are facing a strange issue as we are no more able to re-attach a disk (vRDM type) which has been extended at the Storage level and which is shared between 2 VMs (Cluster RHN). it was previouslay attached to this VM.
To take into account its new size, we dettached it and tried to re-attach it but when we try that, using the "Change Parameters // Add  new disk", we cant see it in the list of candidate disks.

its behavior is the same as if it was not shared at all.

Any help would be greatly appreciated.
Rgds
Hubert

[a_p_]

I assume that the vRDM did/does not have an active snapshot. If it does stop reading at this point now!

If you only detached the vRDM from the VM's without selecting the "Delete from disk" option, the .vmdk RDM mapping files still exist. In order to be able to attach the resized RDM LUN again, you need to delete these mapping files from the datastore. As long as they exist, you can only re-attach the vRDM "as an existing disk", which however still has the old size, and geometry information in its descriptor file.

André

[hpoirier8]

Hi,

Thanks for your reply.

Well, after checking on the GUI interface, browsing the datastore containing the different VMDK files, we noticed, as you explained that the file (pointer file) describing the vRDM file is still existing.
if we checked thru the SSH interface on the ESXi hosting the VM, we can see the specific VMDK file along with the xxx.RDM.vmdk file associated.
We moved these 2 files in a "save" folder , then tried to add a new RDM disk at the VM level (without stopping the VM), but the LUN didn't appear as a candidate LUN/disk. Any explanations?

We also tried to add a new disk, using the "Add Existing disk" option, and choose the specific file (browsing the different folders) but we got the following error :

"Failed to lock the file
Cannot open the disk '/vmfs/volumes/5a76d0ff-29374c60-0da2-ecb1d7b559f0/sli-strsrv-02-prd-cit/sli-strsrv-02-prd-cit_3.vmdk' or one of the snapshot disks it depends on.
Failed to power on scsi1:1.
Failed to add disk scsi1:1"

For your information,
1. No existing snapshot on this VM.
2. All operations done with VM on and running.

Any help would be greatly appreciated.

Thanks in  advance .

[a_p_]

The reason for the LUN being filtered is that you did not delete these files.
The error message that you received with the "Add Existing disk" option is most likely cause by the movement into another folder.

Note that deleting the mapping files for a RDM (unless it has snapshots - that's why I've asked) does not affect the data on the LUN itself.

André

[hpoirier8]

Hi,

thanks for your update.

We tried different actions using the Option "Add Existing disk" but still not working .

the last ones :  VM power OFF : We copied the 2 files in a backup folder , then we add the disk, using the option "Add Existing disk" , it worked but impossible to start the VM (File Lock).

We investigated and noticed that the lock was managed by the VM we are working on.
So we decided to removed the disk from the VM, with the option 'Remove disk from dsik Array" checked. The disks has been removed from the folder. Then we moved the Files backuped (vmdk & rdm.vmdk) in the correct folder (VM folder) and tried to attached them again using the same option "Add Existing disk" but the disk are not in the list, when browsing the datastore Folder ,

Now we are in the state where the files (vmdk & rdm.vmdk) are present in the Datastore (we can list them using the SSH client on the ESXi hosting the VM, in the correct folder), but are no more visible by the VM when we try to add it using the "Add Existing disk" option.
We dont know hat is the issue .
Thanks for your help

Hubert

[a_p_]

Why don't you delete these files?
Once the files are gone, the RDM LUN should be visible again when you add it to a VM as a RDM. Adding the LUN as a RDM will then create the mapping files again with the correct settings.

André