BTRFS pool crashing issues

[greyday]

I have three pools that I use for shares for Machinaris/Chia mining that are made up of 5 old laptop drives each. Two of them work fine, however one of them has a drive that for whatever reason will drop out of the pool after a day or two of use. The drive is fine, this is most likely an issue with the hardware I am using to connect it and I am working on that, however I have a bigger problem: Unraid can add the drive back into the pool without a problem, but when I start the array up again that pool reads as "Unmountable: No file system". I tried the cli suggestions in the FAQ and none of them worked; the first time I figured it may have just been some kind of fluke so I just reformatted. Everything worked fine until the drive dropped out again, and same thing again.

 

Before erasing all the work on it I wanted to see if anyone has had this problem before? The drives are set up as JBOD (or whatever BTRFS calls it, can't remember offhand) so theoretically even if that drive failed the data on the other four SHOULD be fine, right? Is there a way I can access it?

[JorgeB]

On 3/18/2022 at 11:46 PM, greyday said:

so theoretically even if that drive failed the data on the other four SHOULD be fine, right?

No, a non redundant btrfs pool won't mount with a missing device, if you can add the device back it should mount, but you need to be careful how you add the device back, you need to make Unraid forget the pool, or it will wipe the added device, I can post the instructions if still needed.

[greyday]

Thank you. If you have the time at some point the directions would be great (especially for someone else who might stumble on this thread with a similar problem). Since I already tried to add the drive back into the pool I’ll likely not bother trying anything other than solving the hardware problem and rebuilding the plots, but it would definitely be helpful for future reference. I don’t quite get why btrfs non-redundant pools aren’t like JBODs, so they’re basically raid0 without the added speed?

 

EDIT: leaving a note for myself to avoid bumping the thread: if unRAID assigns a new ID to one of the drives, remove all drives from pool, reboot, then check to make sure the ID is back to normal and reassign, then start array. It will take a couple minutes but the pool will restore.

Edited April 6, 2022 by greyday

[JorgeB]

35 minutes ago, greyday said:

I don’t quite get why btrfs non-redundant pools aren’t like JBODs

With the single profile in btrfs data is written in 1GiB chunks always to the device with most available space, so if or when they have similar available space it will alternate between devices, so a single >1GiB file will always be distributed among them.

 

37 minutes ago, greyday said:

so they’re basically raid0 without the added speed?

Basically yes, might as well use raid0, unless using different size devices.

 

37 minutes ago, greyday said:

If you have the time at some point the directions would be great

Basically what you can't do is add the dropped device if Unraid considers that a new device, since it will be wiped and without it the pool won't mount again, when Unraid thinks it's a new device an "all data on this device will be deleted" warning will appear in front of it, to properly add it you can:

 

-stop the array

-unassign all members from that pool

-start the array so the pool config is "forgotten"

-stop array

-now re-assign all existing pool devices including the one that dropped earlier

-there can't be an "all data on this device will be deleted" warning for any of the devices

-start the array and Unraid will import the existing pool