Open
Description
the problem:
# btrfs fi balance cancel /d
ERROR: balance cancel on '/d' failed: Not in progress
# btrfs balance start -mconvert=dup,soft /d/
ERROR: unable to start balance, another exclusive operation 'balance' in progress
some info:
# lsblk
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINTS
sde 8:64 0 12.7T 0 disk /d
# lsblk -z
NAME ZONED ZONE-SZ ZONE-NR ZONE-AMAX ZONE-OMAX ZONE-APP ZONE-WGRAN
sde host-managed 256M 52156 0 128 672K 4K
# dmesg -T | tail -n3; date
[Thu Aug 8 22:50:44 2024] BTRFS info (device sde): found 1330 extents, stage: move data extents
[Thu Aug 8 22:51:02 2024] BTRFS info (device sde): reclaiming chunk 73810086723584 with 6% used 93% unusable
[Thu Aug 8 22:51:02 2024] BTRFS info (device sde): relocating block group 73810086723584 flags metadata|dup
Thu Aug 8 22:49:48 CST 2024
# cat /sys/fs/btrfs/8d4bcd03-3f85-4aae-a7dc-a302f1d6d8bb/allocation/data/bg_reclaim_threshold
75
I'm not sure but there's no btrfs-progs doing balance by checking "ps -efH | grep balance", but dmesg shows the reclaim is in progress. I'd guess it was triggered by bg_reclaim_threshold. I suppose it can be canceled but it's not recognized by btrfs-progs
btrfs --version
btrfs-progs v6.10
reproducible with both v6.10 v6.9.2
uname -a
Linux XueferDG 6.9.9-gentoo #1 SMP PREEMPT_DYNAMIC Tue Jul 16 02:05:40 CST 2024 x86_64 Intel(R) Xeon(R) CPU D-1581 @ 1.80GHz GenuineIntel GNU/Linux