Hey.

My server come up and had a missing btrfs device (diagnostics) attached.

I shut down, checked cables and the device returned. Now it is no longer part of the cache pool.

How do I get it back in?...

knowlage-diagnostics-20190910-1524.zip

Diagnostics after reboot:

knowlage-diagnostics-20190910-1540.zip

The device is already part of the pool, since it's available again, despite not being assigned, to correct the assignments it's easier to unassign all cache devices (stop VM and docker services first), then start array without any assigned cache devices so Unraid can "forget" current config, then stop the array and re-assign all cache devices, you should then run a scrub to bring the offline device up to date.

Cool. Thanks. Back onlline and scrubbing....

 

 

How to deal with the resulting uncorrectables?

scrub status for 841ae120-6e82-44b2-bd5f-ee9fd1de1907
	scrub started at Tue Sep 10 17:55:12 2019, running for 00:00:41
	total bytes scrubbed: 9.82GiB with 2595 errors
	error details: read=4 verify=2571 csum=20
	corrected errors: 2588, uncorrectable errors: 7, unverified errors: 0

With uncorrectable errors best way forward is to backup any important data on cache and reformat the pool.

Any other way? As the pool only contains VMs and docker data..?

Uncorrectable errors can be the result of using NOCOW shares, more info here, you need to delete affected data or format pool, no other way.

I dont have any nocow shares. But it seems I cant backup anything ->

cp: error reading 'vdisk1.img': Input/output erro

Is there any way to save this? It is vital for me as I would lose some veryyyy imporant data is just received today (nightly backup did not run yet)

 

It is a RAID6, how can it fail in a way I can not recover anything!? 

 

[/dev/sdm1].write_io_errs 0
[/dev/sdm1].read_io_errs 0
[/dev/sdm1].flush_io_errs 0
[/dev/sdm1].corruption_errs 0
[/dev/sdm1].generation_errs 0
[/dev/sdh1].write_io_errs 0
[/dev/sdh1].read_io_errs 0
[/dev/sdh1].flush_io_errs 0
[/dev/sdh1].corruption_errs 93230
[/dev/sdh1].generation_errs 2571
[/dev/sdk1].write_io_errs 0
[/dev/sdk1].read_io_errs 148
[/dev/sdk1].flush_io_errs 0
[/dev/sdk1].corruption_errs 0
[/dev/sdk1].generation_errs 0
[/dev/sdj1].write_io_errs 0
[/dev/sdj1].read_io_errs 0
[/dev/sdj1].flush_io_errs 0
[/dev/sdj1].corruption_errs 0
[/dev/sdj1].generation_errs 0

Why does the cache appear to be read only?!

Edited September 10, 20196 yr by Jaster

btrfs raid5/6 still has some issues, there's a warning in the FAQ, input/output error when copying means the file is corrupt, you can override that by using btrfs restore, but the file will still be corrupt, it might be justle a little in a place that doesn't matter, or it might matter, but you can try.

 

Anything important needs to be backed up, even when using other raid profiles or other filesystem.