Cache drives in unassigned devices?

[hermy65]

I moved my server to a new location, fired it up and for some reason 2 of my cache drives are now showing up in unassigned devices. If i add them i assume it rebuilds them but do i lose any data? Whats the best way to handle this so i dont lose anything on my cache drive? Also, why would this happen?

[John_M]

7 minutes ago, hermy65 said:

I moved my server to a new location

That is probably the clue. The SATA and/or power cables probably got disturbed. Samsung SSDs are incredibly fussy about their SATA cables at the best of time.

[hermy65]

5 minutes ago, John_M said:

That is probably the clue. The SATA and/or power cables probably got disturbed. Samsung SSDs are incredibly fussy about their SATA cables at the best of time.

@John_M so do I just add them back into the cache pool then? Will I lose any data?

[John_M]

I'd power down and check the cables first. Then power up and reassign the two SSDs, if necessary. btrfs should run a balance when you start the array.

[John_M]

Losing two devices out of four is serious but if you didn't start the array in that condition I would not expect you to lose any data, but one can never be sure. What RAID level are you running - I assume 1 or 10?

[hermy65]

@John_M It was RAID 1. I checked the cables and they all seem solid. I added the two drives back into the pool and started the array but now im getting this.

 

[JorgeB]

5 hours ago, hermy65 said:

I added the two drives back into the pool and started the array but now im getting this.

You'll destroy the pool doing that, for the future the correct procedure would be:

 

Stop the array, if Docker/VM services are using the cache pool disable them, unassign all cache devices, start array to make Unraid "forget" current cache config, stop array, reassign all cache devices, re-enable Docker/VMs if needed, start array.

 

Alternatively you can also do a new config and reassign all devices, then check parity is already valid before starting the array.

[hermy65]

@johnnie.black I followed your steps but when i re-add the drives and start the array i get an error that says: Cache drive unmountable, no pool uuid.

 

I tried the btrfs restore steps listed here and was able to recover data from the two drives in my original screenshot that were in unassigned devices but when i try it on the two disks that were listed in the pool in my original picture i get the error in the attached image. 

 

[JorgeB]

38 minutes ago, hermy65 said:

I followed your steps but when i re-add the drives and start the array i get an error that says: Cache drive unmountable, no pool uuid.

Like mentioned:

5 hours ago, johnnie.black said:

You'll destroy the pool doing that

 

[hermy65]

@johnnie.black I guess im not following. Are you saying what i did earlier caused the issue? 

[JorgeB]

Just now, hermy65 said:

Are you saying what i did earlier caused the issue? 

Yes, you can't re-add existing devices like that, there would be a "any data on this devices will be lost" warning next to them, and starting the array wipes them, a 4 disk pool can't mount with 2 missing devices, the correct way of fixing the pool would be what I wrote above.

[John_M]

I gave bad advice. I'm mortified.

[JorgeB]

It will happen to anyone sooner or later, cache pool behavior especially can be very tricky sometimes, also a reminder that everyone should keep up to date backups of anything important.

 

 

[albion01]

On 5/8/2020 at 2:03 AM, JorgeB said:

You'll destroy the pool doing that, for the future the correct procedure would be:

 

Stop the array, if Docker/VM services are using the cache pool disable them, unassign all cache devices, start array to make Unraid "forget" current cache config, stop array, reassign all cache devices, re-enable Docker/VMs if needed, start array.

 

Alternatively you can also do a new config and reassign all devices, then check parity is already valid before starting the array.

 

Lost a power supply, after replacing the unit my raid came up with one of the Cache Pool drives unassigned. This solution worked great to get my dockers and VMs back.

 

Thanks!