Hi Paul, Thank you for the ping, much appreciated :-) Theodore, sorry for needing to CC you, I took a month to try to find the answer on my own, but was unsuccessful. Please see below for the question apropos ext4+cgroupv2+iogroups. If you know if other filesystems are supported, please let us know! Paul Gevers <elbrus@debian.org> writes: > Hi Nicholas, > > On 17-04-2021 00:35, Nicholas D Steeves wrote: >> Last I checked, only btrfs supports the cgroupv2 I/O controller; this >> should probably be documented. Alternatively, if more than btrfs (ie: >> XFS and ext4) supports it, but not other file systems (ie: f2fs, >> reiser4, jfs, etc) than this should be documented. >> >> As far as I can tell, the significance of this is as follows: Buster >> switched to mq-deadline, which does not support the idle I/O priority, >> and to get functional ionice the user/sysadmin had to switch to the >> non-mq CFQ, or bfq. The situation in a default Bullseye install is >> better, but only for file systems that support the cgroupv2 I/O >> controller. It seems I sent the correction to the wrong email. From a second round of investigation, it appears that the cgroupv2 iogroups also require either CFQ (non-mq) or BFQ (mq). > > As I don't have much knowledge on this front, can you maybe do an actual > proposal for text to be added? And, if you're unsure, how do we find out > if you're right? > I'm also not sure, because everywhere I've looked appears to document that this technology is not filesystem specific (except the Facebook cgroupv2 iogroup announcement which asserts btrfs-only). CCing Theodore Ts'o, who will definitely know if ext4 is supported! Regards, Nicholas
Attachment:
signature.asc
Description: PGP signature