bcachefs
bcachefs copied to clipboard
koverstreet
Handle metadata in rebalance
After adding a new disk to my fs, with label hdd, and the filesystem being formatted with metadata_target=ssd, somehow metadata ended up on the new disk (will report that as a separate issue). See bcachefs fs usage, the new drive is sdl:
Filesystem: b402b3da-8057-4d2f-acdc-4a3de16a7c38
Size: 162 TiB
Used: 153 TiB
Online reserved: 15.5 MiB
Data type Required/total Durability Devices
hdd (device 4): sdb rw
data buckets fragmented
free: 332 GiB 339865
sb: 3.00 MiB 4 1020 KiB
journal: 8.00 GiB 8192
btree: 0 B 0
user: 10.5 TiB 11096163 33.1 GiB
cached: 0 B 0
parity: 0 B 0
stripe: 0 B 0
need_gc_gens: 0 B 0
need_discard: 0 B 0
unstriped: 0 B 0
capacity: 10.9 TiB 11444224
hdd (device 3): sdc rw
data buckets fragmented
free: 332 GiB 340002
sb: 3.00 MiB 4 1020 KiB
journal: 8.00 GiB 8192
btree: 0 B 0
user: 10.5 TiB 11096026 34.6 GiB
cached: 0 B 0
parity: 0 B 0
stripe: 0 B 0
need_gc_gens: 0 B 0
need_discard: 0 B 0
unstriped: 0 B 0
capacity: 10.9 TiB 11444224
hdd (device 11): sdd rw
data buckets fragmented
free: 352 GiB 360728
sb: 3.00 MiB 4 1020 KiB
journal: 8.00 GiB 8192
btree: 0 B 0
user: 10.4 TiB 11075300 143 GiB
cached: 0 B 0
parity: 0 B 0
stripe: 0 B 0
need_gc_gens: 0 B 0
need_discard: 0 B 0
unstriped: 0 B 0
capacity: 10.9 TiB 11444224
hdd (device 6): sde rw
data buckets fragmented
free: 606 GiB 620686
sb: 3.00 MiB 4 1020 KiB
journal: 8.00 GiB 8192
btree: 0 B 0
user: 19.3 TiB 20352878 77.7 GiB
cached: 0 B 0
parity: 0 B 0
stripe: 0 B 0
need_gc_gens: 0 B 0
need_discard: 0 B 0
unstriped: 0 B 0
capacity: 20.0 TiB 20981760
hdd (device 7): sdf rw
data buckets fragmented
free: 606 GiB 620547
sb: 3.00 MiB 4 1020 KiB
journal: 8.00 GiB 8192
btree: 0 B 0
user: 19.3 TiB 20353017 77.2 GiB
cached: 0 B 0
parity: 0 B 0
stripe: 0 B 0
need_gc_gens: 0 B 0
need_discard: 0 B 0
unstriped: 0 B 0
capacity: 20.0 TiB 20981760
hdd (device 5): sdg rw
data buckets fragmented
free: 605 GiB 619981
sb: 3.00 MiB 4 1020 KiB
journal: 8.00 GiB 8192
btree: 0 B 0
user: 19.3 TiB 20353583 62.2 GiB
cached: 0 B 0
parity: 0 B 0
stripe: 0 B 0
need_gc_gens: 0 B 0
need_discard: 0 B 0
unstriped: 0 B 0
capacity: 20.0 TiB 20981760
hdd (device 9): sdh rw
data buckets fragmented
free: 612 GiB 626631
sb: 3.00 MiB 4 1020 KiB
journal: 8.00 GiB 8192
btree: 0 B 0
user: 19.3 TiB 20346933 110 GiB
cached: 0 B 0
parity: 0 B 0
stripe: 0 B 0
need_gc_gens: 0 B 0
need_discard: 0 B 0
unstriped: 0 B 0
capacity: 20.0 TiB 20981760
hdd (device 8): sdi rw
data buckets fragmented
free: 606 GiB 620688
sb: 3.00 MiB 4 1020 KiB
journal: 8.00 GiB 8192
btree: 0 B 0
user: 19.3 TiB 20352876 76.7 GiB
cached: 0 B 0
parity: 0 B 0
stripe: 0 B 0
need_gc_gens: 0 B 0
need_discard: 0 B 0
unstriped: 0 B 0
capacity: 20.0 TiB 20981760
hdd (device 10): sdj rw
data buckets fragmented
free: 629 GiB 643825
sb: 3.00 MiB 4 1020 KiB
journal: 8.00 GiB 8192
btree: 0 B 0
user: 19.3 TiB 20329739 111 GiB
cached: 0 B 0
parity: 0 B 0
stripe: 0 B 0
need_gc_gens: 0 B 0
need_discard: 0 B 0
unstriped: 0 B 0
capacity: 20.0 TiB 20981760
hdd (device 12): sdl rw
data buckets fragmented
free: 14.8 TiB 15500007
sb: 3.00 MiB 4 1020 KiB
journal: 8.00 GiB 8192
btree: 32.8 GiB 116250 80.7 GiB
user: 5.03 TiB 5341519 70.1 GiB
cached: 275 MiB 15788 15.1 GiB
parity: 0 B 0
stripe: 0 B 0
need_gc_gens: 0 B 0
need_discard: 0 B 0
unstriped: 0 B 0
capacity: 20.0 TiB 20981760
ssd (device 1): sdk4 rw
data buckets fragmented
free: 207 GiB 211629
sb: 3.00 MiB 4 1020 KiB
journal: 8.00 GiB 8192
btree: 379 GiB 684132 289 GiB
user: 40.0 GiB 49458 8.30 GiB
cached: 204 GiB 619445 401 GiB
parity: 0 B 0
stripe: 0 B 0
need_gc_gens: 0 B 0
need_discard: 4.00 MiB 4
unstriped: 0 B 0
capacity: 1.50 TiB 1572864
ssd (device 0): sdm4 rw
data buckets fragmented
free: 207 GiB 212024
sb: 3.00 MiB 4 1020 KiB
journal: 8.00 GiB 8192
btree: 379 GiB 683830 289 GiB
user: 40.0 GiB 49469 8.31 GiB
cached: 203 GiB 619340 402 GiB
parity: 0 B 0
stripe: 0 B 0
need_gc_gens: 0 B 0
need_discard: 5.00 MiB 5
unstriped: 0 B 0
capacity: 1.50 TiB 1572864
Dumping some snippets from IRC:
23|17:59:35 < stintel> 23|09:17:00 <@py1hon> there ought to be an easy way to move data/metadata to the correct target
23|17:59:40 < stintel> any pointers where to look?
23|17:59:46 <@py1hon> no
23|17:59:55 <@py1hon> "ought to be", not "is" :p
23|17:59:59 < stintel> I see
23|18:00:43 <@py1hon> technically that should be rebalance's job
23|18:00:59 <@py1hon> but rebalance doesn't know anything about moving metadata around, just data
23|18:02:09 <@py1hon> we'd want to add the concept of a rebalance scan for metadata, not data
23|18:02:39 <@py1hon> scans get triggered when you flip a filesystem level or inode option, they're cookies in the rebalance work btree
23|18:02:46 <@py1hon> so add a new cookie type
