News for hackers. Stuff that matters.
Re:Linux 5.10 BTRFS performance regression by StavrosK on Wednesday December 23, 2020 @02:50PM (#25521404)
It's mostly sequential performance, but it's also very odd. Deleting a single 2 GB file sometimes takes tens of seconds, or copying files from one volume to another will again be extremely slow.. .
Re:Linux 5.10 BTRFS performance regression by magicalhippo on Thursday December 24, 2020 @03:25AM (#25525927)
It does sound like fragmentation, which can still happen with a pool that's not filled under the "right" circumstances (usually related to workload vs configuration).

But yeah sadly that's the one area where ZFS is less stellar. Once it's fragmented it's hard to fix. Easiest is to send/receive to another pool, but as you note that is not always feasible.

Re:Linux 5.10 BTRFS performance regression by StavrosK on Thursday December 24, 2020 @05:33AM (#25526432)
Is there no defragmentation tool? I do have lots of free space.
Re:Linux 5.10 BTRFS performance regression by magicalhippo on Thursday December 24, 2020 @05:12PM (#25531952)
Sadly no, not really. Part of it is that it's complicated due to copy-on-write and snapshots. Other part is manpower, it has not been a high-priority issue for most production pools it seems. Google for "block pointer rewrite".

Assuming your free space fragmentation is not too bad (check output of zpool list, "frag" column is free space fragmentation level), you could just move files back and forth between the two. Assuming you don't have snapshots holding on to the files, this can help reduce the fragmentation.

Otherwise you're stuck with the send/receive.

Though I'd try the mailing list[1] to see if any of the gurus can help identify what's going wrong before attempting random ailments.

[1]: https://zfsonlinux.topicbox.com/groups/zfs-discuss/

Re:Linux 5.10 BTRFS performance regression by StavrosK on Thursday December 24, 2020 @05:15PM (#25531981)
Hmm, I'll try the mailing list, thank you for your help!