I have some virtual machines that are using RDM disks as RawPhysical. When my vendor increased the disk space for the RDM disks, it only reflect on OS level. (Windows Server) However, the new LUN size doesn't reflect accordingly on VMHost level (Get-VMHost $VMHost | Get-ScsiLun -CanonicalName $RDM). And also not reflected on VM level when using (Get-Harddisk -VM $VMName). I have tried rescanning all hba and reloading the VM but still with wrong LUN size.
Have a look at KB2114219
Blog: lucd.info Twitter: @LucD22 Co-author PowerCLI Reference
I have found even after doing that KB procedure if you query the size using powercli for reporting (which we do as a company) it still shows the old size, the only way we have found to make it work correctly is to recreate the RDM pointer when increasing the size.
This is quite an annoying process at least for us, pretty time consuming to gather all the information for verification and recreation of the pointer exactly as it was before on the same ID/location so the cluster can just carry on as normal.
Happy to be corrected though if there is a better way.
Yep, I've actually tested the KB above. The only way to get the correct LUN sizing is by detaching the RDM from ESXi Host Layer and attaching them back. That way the new size will reflect accordingly, however, this way, the VM using the LUN will also need to undergo disk removal process.
This is a known issue with the RDM mapping file. See also RDM LUN size mismatch
When a LUN is extended, this file (example-rdm.vmdk) is not updated.
And PowerCLI somehow seems to use this to find the size (same as RVTools btw).
On the other hand the Web Client seems to list the correct size (and so is the guest OS).
I was wondering if the correct size is shown with for example
Blog: lucd.info Twitter: @LucD22 Co-author PowerCLI Reference
I will give that a go in lab test, maybe easier to change the reporting method than have to go through removing and adding the vmdk each time.