On Saturday, May 22, 2021, 03:14:13 PM GMT+8, Eugen Block <eblock(a)nde.ag> wrote:
What does the MDS report in its logs from when it went down?
What size do you get when you run
rados -p cephfs_metadata stat 200.00006048
# rados -p cephfs_metadata stat 200.00006048cephfs_metadata/200.00006048 mtime 2021-05-20
22:47:30.000000, size 1555896
There's a similar report [3] suggesting to try to force an update on
the object info, you could give that a shot:
1. rados -p [cephfs_metadata] setomapval 200.00006048
temporary-key anything
2. ceph pg deep-scrub 2.44
3. Wait for the scrub to finish
4. rados -p [cephfs_metadata] rmomapkey 200.00006048 temporary-key
I gave it a try. Here are the detail:
Before try it:rados list-inconsistent-obj 2.44 --format=json-pretty{ "epoch":
6996, "inconsistents": []}
After trying the step 1:# rados list-inconsistent-obj 2.44 --format=json-pretty{
"epoch": 6996, "inconsistents": [ {
"object": { "name": "200.00006048",
"nspace": "", "locator": "",
"snap": "head", "version": 0
}, "errors": [], "union_shard_errors":
[ "obj_size_info_mismatch" ],
"shards": [::
That is, It created the info for PG 2.44.
It shows size differences as follows:shards": [ {
"osd": 0, "primary": true,
"errors": [ "obj_size_info_mismatch"
], "size": 1540096,
"object_info": { "size":
1555896,
"osd": 1, "primary": false,
"errors": [ "obj_size_info_mismatch"
], "size": 1540096,
"object_info": { "size": 1555896,
"osd": 2, "primary": false,
"errors": [
"obj_size_info_mismatch" ],
"size": 1441792, "object_info": {
"size": 1555896,
Please note, the physical file size of 200.00006048 on OSD 0 and 1 is 1540096.
Physical file size on OSD 2 is 1441792.
I understand file sizes of the 200.00006048 should be same on all OSDs.
What should I do in that case?
Please also note, "ceph pg deep-scrub 2.44" did not fix the PG 2.44.
Sagara