VMware

Upgrading disks in-place, ideally while running

I’ve got a slightly older machine running ESXI 5, and I’m looking to upgrade the drives in it – it’s got a pair of 100GB drives and 200GB drives, both mirrored. I’m hoping I can upgrade them in-place with no downtime by simply removing one of the parity drives, replacing it with a larger drive (maybe 4TB?) and waiting for the host to rebuild the mirror on the new drive. Is this possible (and is it a safe idea)? Additionally, is there a way I can verify that the drives are indeed mirrored before I start yanking them? (I got this machine second hand, so I’d like to double check the configuration before I go ahead and depend on it)


View Reddit by iliketanks1998View Source

Related Articles

4 Comments

  1. What RAID controller are you using?

    You should check to see if the controller supports hot expansion of the RAID virtual disk even if you replace with larger disks.

    TBH I don’t recall ever doing this with RAID-1; only with parity arrays and that was with adding additional disks.

  2. > I’m hoping I can upgrade them in-place with no downtime by simply removing one of the parity drives, replacing it with a larger drive (maybe 4TB?) and waiting for the host to rebuild the mirror on the new drive.

    The mirror will only rebuild to the size of the smallest disk in the array – so your 4TB drive won’t give you 4TB. It may be possible to expand the mirror if the RAID controller supports online expansion of the virtual disk/array once both disks are replaced and the extra 3.9TB is available on both disks.

  3. For shits’n’giggles, I totally would love to see a successful “process” of this working.

    But! Murphy’s Law will apply here, and I suggest at best. Backing up those RAID Arrays to a external HDD. You should perform a Disk to Image backup (get your existing partitions whatever, just backup).

    Try the in place HDD swap. Let the array rebuild, and then finish the next drive. (for both the 100GB and 200GB drives). Afterwards, you may be able to extend the filesystems(volumes).

    However all of this depends on the hardware you have running this setup. “older machine” just tells me a “white box” with a single Xeon, and the RAID is onboard. Versus a Dell PE R410 “older machine” 2x Xeon, with a H710 PERC controller. I almost fsck’d a Dell T420, by accidently popping out a HDD caddy with an actual running 1TB drive. While it was running. Having good backups, just helps secure from the accidental loss. If you have your backup ready/tested. Then as you try your untested “idea”, your odds for success increase. Even if you fail, you still have backups.

  4. plan a downtime, reboot that server and look at the raid config (don’t do changes)

    on raid1 if you replace with a larger disk the size of the volume will remain the same and the extra free space won’t be allocated

    my rule for 2nd hand servers is to always put new discs in, even if the old ones are still good

    so, based on the above:

    – backup/export/migrate/copy your vm’s (whatever option works for you)
    – get 4 new identical disks and make a raid 10 if available on the controller or 2 raid 1 if not
    – clean esxi install on the new disks
    – restore the vm’s

    LE – regarding the esxi install, many servers have a usb/sd card port inside the case and you could use a usb stick or sd card for boot and leave the raid volumes for the datastore only

Leave a Reply

Your email address will not be published. Required fields are marked *

Close