Re: qgroup rescan worker makes suspend fail

[Qu Wenruo <wqu@suse.com>]

在 2026/9/20 14:26, Andrei Borzenkov 写道:
> 20.09.2026 01:39, Qu Wenruo wrote:
>>
>> To be honest, qgroup mode is no longer recommended, except for rigid
>> subvolume layouts, and since you need rescan it's definitely the not
>> recommended case.
>>
>> The current only well known user is snapper, and we're pushing snapper
>> not to utilize qgroup by default.
>>
>> So unless you have a very clear use case, it's better just disable
>> qgroups completely.
> 
> I am not aware of any other way to get true subvolume space consumption 
> and thus implementing space based retention policy. "btrfs du" is hardly 
> an option.
> 
> If qgroup is deprecated, what are other options to get this information?

Either simple mode, or really "btrfs fi du".
Or in the future, regular quota file support.

But neither will provide the current accurate "exclusive" accounting.
Simple mode is the closest one to provide exclusive result with the 
minimal overhead, but very un-instinctive values for deleted subvolumes.

"btrfs fi du" is also fine, but requires execution before getting the 
result, and can be very slow for a lot of small files.
At least it won't block other btrfs operations, and can properly handle 
pm suspension.

Future quota file support can only provide "reference" level accounting, 
but should have minimal overhead and should require no frequent rescan 
(except the initial enabling), but no "exclusive" accounting at all.