19/11/2017 - The Plan

Posted on Sun 19 November 2017 in Blog

"I now have to re-address my backups and sort out the performance issues which could lead to more data loss! I need at least another week…."

I've experimented with a spare SDD. I can get a working passthrough desktop on it. Great, I can reduce the load on DS01. The really old 1TB disk. See the problem? I'm only fixing half of the issue and making a future fix impossible. * Really, really hard and possibly expensive. DS01 is old. It's time to be replaced, before it fails, taking my entire infrastructure with it.

I have eight SATA ports with which to work. Five are taken by large drives. One is taken by a cloud backup drive. One is DS01 and one is currently spare. If I use the spare port for the desktop SSD now, I have no free ports for migrating DS01 later. I need move the desktop to a different disk ASAP. I need to migrate DS01 first.

After deliberation, I've purchased a 1TB SSHD. It was almost a 500GB SSD but I feel I'm getting better speed vs storage vs price. 500GB could disappear quickly. 1TB is far less likely to. Once I migrate DS01 to the new disk and remove DS01, I'll still have a free port for my SDD desktop.

Migrating DS01 I still need to do some reading and would like to do another few tests however I believe there are two options for me. One: Remove DS01, clone it to the new SSHD. Untested, most likely will work. It's a D2D clone. Two: You can just move stuff in ESXi. I moved a test VM on the SSD to DS01, it took an hour or so. It worked. A VM is just a .vmdk file with a bunch of smaller config files. An RDM is just a .vmdk file. I'm going to work some magic, move test VM back over to the SSD, attach a secondary drive as MK8 has, migrate the test VM back over to DS01 then move the .vmdk drive and see what happens. This https://kb.vmware.com/s/article/1005241 says it's ok to do.

The RDM is just a pointer file. It has nothing to do with the data on the disk.

I quote • When performing a cold migration of a virtual machine with RDMs attached to it, the contents of the raw LUN mapped by the RDM are copied into a new .vmdk file at the destination, effectively converting or cloning a raw LUN into a virtual disk. This also applies when the virtual machine is not moving between ESX hosts. In this process, your original raw LUN is left intact. However, the virtual machine no longer reads or writes to it. Instead, the newly-created virtual disk is used.

It should be simple enough to accomplish. He says.

The is also the underlying backup issue to address. Two of my failed drives are under warranty so can be renewed. That'll take a week or so to happen. I don't have enough working disks to fully backup. I'll need to take desperate measures and selectively backup. I would rather, as I have had every single time I've done this, have cold data. Unfortunately, given the risk of impending disaster, selective will have to do this time. Drive pool should have my back. All the new disks are ok. SSHD arrives sometime tomorrow, we'll see….