This is the continuation and Part 8 of my Storage Spaces and Data Deduplication review. Here’s an index of test cases on this part and links to other parts:
- Part 1: Introduction and Lab Environment Preparation
- Part 2
- Physical Disk Pull
- Introduce the Pulled Disk Back into the System
- Part 3
- Extend Thinly Provisioned Virtual Disk and Observe Behavior at Limits
- Bonus: Detecting and Replacing Physical Disk Failures
- Part 4
- Removing a Disk from the Storage Pool
- Part 5
- Reclaim Unused Space on Thin Provisioned Disks
- Bonus: Defragmentation Attempt and Observations
- Part 6
- Understanding Hot-Spare Behavior
- Part 7
- Part 8 (You’re here)
- Repurposing Disks from other Servers
Repurposing Disks from other Servers
In my continued attempt to discover intricate operational details of Storage Spaces, I wanted to experiment with shuffling disks between systems and see what behavior I get.
Here’s how things are now:
I have performed the following:
Action: Remove 92GB drive. Insert 1 x 1.5TB, 1 x 750GB and 1 x 2TB disks (total 3). Keep in mind, the 92GB disk was part of the TestSP1 pool that is parity protected. Here’s the result:
What’s interesting above is that I do not see my three new disks. They were dynamic disks and were in use in another Windows Server prior to bringing here. Health degradation is expected of course, since one of the disks in the pool with parity protection is now gone.
I also see heavy disk activity. I can’t explain what this is – for now.
One idea maybe that system is “healing” the virtual disks by re-generating the parity that was lost on the missing drive. However, status of virtual disks are not indicating any repair in progress. As in:
How about physical disks? The UI shows us this:
Where are those three disks that I have added to the system? Let’s try Powershell instead:
What do you know… They do show up in “get-physicalDisk” command. Strange.
What’s even more strange is the duplications in FriendlyName values. If you note in above screenshot, those showing with 698GB, 1.36TB and 1.82TB are the disks that I have just added. They physicalDisks 2, 4 and 5. But look at the 2nd line. The disk with 278.75GB capacity is also called PhysicalDisk5. I will investigate what’s going on here some more.
Let’s look at Disk Management UI and see what it’s showing:
In above screenshot; finally I can see the disks I have just added. They arrive in offline mode.
Only one way to find out what’s in them. I will attempt to bring them online. I do this by individually doing a right-click on each disk and choose “Online” command.
Ok, now they’re reporting as foreign but accessible. Let me do “Import Foreign Disk” on one of these disks and see what happens. When I execute that command, following dialog popped up. The dialog contained the volumes that these disks were part of.
It’s also warning me that I had not brought in all the disks that comprised those previous volumes. At this point I don’t care about data loss but this is the experience you’d have if we you were moving a group of dynamic disks to another system. I clicked “Yes” on the dialog above.
Now I see the remnants of the volumes from previous life of these disks. I shall clean them up. Simply right-click on each failed volume and delete.
At this point I should check the PowerShell and see if anything is different.
Sure enough, “CanPool” column value has changed for those three disks. “Pool” name indicates that this might also be reflected in Storage Pools, quick refresh on the Server Manager UI shows following update:
Now we’ve got Storage Spaces to recognize those disks.
So what did we observe?
- If you bring dynamic disks from another system, you do need to use Disk Management console or diskpart to clean them up. Otherwise Storage Spaces management UI doesn’t show anything about them.
- Physical disk names can be duplicate on the same system. This is very strange – not sure what to make of this or what the implications might be.
- When a disk of an active pool is removed, remaining disks are trashing with no end in sight. I can’t explain why this is happening but the activity does NOT appear to be self-healing or hot-spare activation.
At this point I’m very curious about the disk name duplication, I will proceed to add that new disk5 to the pool and see what happens.
The system doesn’t complain. It can happily have same-name disks in the same pool. Difficulty with this behavior is that if you’ve got same-size disks and if they all named the same, how will you, without “disk light enable” support, be able to determine which disk is which? Say you need to replace a failed disk. It’d be interesting exercise to find out which disk is mapped to which physical/actual disk in the enclosure.
So let’s investigate that.
If you right-click on one of the physical disks and choose “Properties”, you get the following dialog:
As you can see this shows the serial number, which is unique for each physical disk. You could read up your physical disk serials in the enclosure and replace the one that matches the failed disk’s serial number. In real world this is probably an offline operation if your gear doesn’t support “Toggle drive light” operation. Reason being that if you start pulling disks while hot, you’ll cause data outage on 2nd disk that you pull in sequence. If you however shutdown the array and then pull one by one and read the serial number, you’ll be able to safely replace it and power the system back on.
That’s it. Hopefully this post helps save you some time on moving disks between systems.
Categories: Computers and Internet