Re: [BUG] unable to mount zoned volume after force shutdown

[Johannes Thumshirn <Johannes.Thumshirn@wdc.com>]

On 09.06.24 12:33, Qu Wenruo wrote:
> 
> 
> 在 2024/6/9 11:53, HAN Yuwei 写道:
>> I can't mount volume on multiple zoned device which data profile is
>> single & metadata profile is raid1.
>> It experienced multiple forced shutdown due to kernel 6.10 can't
>> properly shutdown on loongarch.
>>
>> # dmesg
>>
>> [ 1963.698793] BTRFS info (device sdb): first mount of filesystem
>> b5b2d7d9-9f27-4907-a558-77e8e86df933
>> [ 1963.707801] BTRFS info (device sdb): using crc32c (crc32c-generic)
>> checksum algorithm
>> [ 1963.715597] BTRFS info (device sdb): using free-space-tree
>> [ 1965.492066] BTRFS info (device sdb): host-managed zoned block device
>> /dev/sdb, 52156 zones of 268435456 bytes
>> [ 1966.953590] BTRFS info (device sdb): host-managed zoned block device
>> /dev/sdc, 52156 zones of 268435456 bytes
>> [ 1967.346758] BTRFS info (device sdb): zoned mode enabled with zone
>> size 268435456
>> [ 2026.287356] BTRFS error (device sdb): zoned: write pointer offset
>> mismatch of zones in raid1 profile
>> [ 2026.296445] BTRFS error (device sdb): zoned: failed to load zone info
>> of bg 5399847632896
> 
> My guess is, due to the forced shutdown, btrfs' writer pointer is later
> than the zone write pointer.
> 
> IMHO, we should not error out but mark this zone as read-only and let
> balance to kick in.
> 
> @Johannes and @naohiro, mind to enhance the error handling there?

That's definitely a possibility we need to look into yes.