Regarding changing this bvalue back to the previous default of
2,000,000, how would I go about doing that? I tried following that
SUSE KB article which says to do this:
ceph tell 'osd.*' injectargs --
osd_deep_scrub_large_omap_object_key_threshold=2000000
But while that didn't fail as such, it didn't apply any changes. Is
there a way to apply this on the fly without restarting the cluster?
On Tue, 2023-08-01 at 22:44 +0000, Mark Johnson wrote:
Thanks for that. That's pretty much how I was reading it, but the
text you provided is a lot more explanatory than what I'd managed to
find and makes it a bit clearer. Without going into too much detail,
yes we do have a single user that is used to create multiple a bucket
for each of a multiple tenants on a daily basis. So, we'd be
creating many buckets each day and all owned by the same account.
Therefore, it's quite possible that there could be 400,000 buckets
owned by the one user. I don't know an easy way to get a figure - I
tried a "radosgw bucket stats" output to a file but after about 4
hours it still hadn't returned anything so I gave up.
I have a feeling that we do have a rolling clean out of objects in
these buckets, so we might be only keeping 3 months of data for some
customers, 6 months for others, 12 months for others etc. But, I
think one of our guys mentioned that the cleanup might not be getting
rid of buckets, only the files in them. So, I may have to get our
dev guys to revisit this and see if we can clean up a crapload of
empty buckets.
On Tue, 2023-08-01 at 08:37 +0000, Eugen Block wrote:
Thanks. Just for reference I'm quoting the SUSE doc [1] you mentioned
because it explains what you already summarized:
User indices are not sharded, in other words we store all the keys
of names of buckets under one object. This can cause large objects
to be found. The large object is only accessed in the List All
Buckets S3/Swift API. Unlike bucket indices, the large object is not
exactly in the object IO path. Depending on the use case for so many
buckets, the warning isn't dangerous as the large object is only
used for the List All Buckets API.
The error shows a user has 500K buckets causing the omap issue.
Sharding does not occur at the user level. Bucket indexes are
sharded but buckets per user is not (and usually the default
max_bucket is 1000).
Does this mean that you actually have a user with around 400k
buckets?
If you can't delete unused buckets (you already ruled out creating
multiple users) there's probably no way around increasing the
threshold, I guess. I'm not the biggest RGW expert but we have a few
customers where the threshold was actually increased to the previous
default to get rid of the warning (if other actions were not
possible). So far we didn't get any reports causing any issue at all.
But I'd be curious if the devs or someone with more experience has a
better advice.
[1] https://www.suse.com/support/kb/doc/?id=000019698
Zitat von Mark Johnson
<markj@xxxxxxxxx<mailto:markj@xxxxxxxxx><mailto:markj@xxxxxxxxx<mailt
o:markj@xxxxxxxxx>>>:
Here you go. It doesn't format very well, so I'll summarize what I'm
seeing.
5.c has 78051 OMAP_BYTES and 398 OMAP_KEYS
5.16 has 80186950 OMAP_BYTES and 401505 OMAP_KEYS
The remaining 30 PGS have zero of both. However, the BYTES for each
PG
is very much the same at around 8900000 for each.
# ceph pg ls-by-pool default.rgw.meta
PG OBJECTS DEGRADED MISPLACED UNFOUND BYTES OMAP_BYTES*
OMAP_KEYS* LOG STATE SINCE VERSION
REPORTED UP ACTING SCRUB_STAMP
DEEP_SCRUB_STAMP LAST_SCRUB_DURATION
SCRUB_SCHEDULING
5.0 26240 0 0 0 8909864 0
0 10076 active+clean 10h 8093'54176
8093:5396520 [21,4,12]p21 [21,4,12]p21 2023-07-
31T21:13:20.554485+0000 2023-07-26T03:40:27.457946+0000
5 periodic scrub scheduled @ 2023-08-01T23:55:14.134653+0000
5.1 26065 0 0 0 8840849 0
0 10029 active+clean 10h 8093'56529
8093:4891333 [14,7,23]p14 [14,7,23]p14 2023-07-
31T20:37:34.920128+0000 2023-07-30T10:55:16.529046+0000
5 periodic scrub scheduled @ 2023-08-01T21:12:04.440688+0000
5.2 26406 0 0 0 8943783 0
0 10076 active+clean 20h 8093'56776
8093:5022283 [26,8,25]p26 [26,8,25]p26 2023-07-
31T11:08:00.886979+0000 2023-07-30T06:03:44.341435+0000
5 periodic scrub scheduled @ 2023-08-01T16:02:32.076634+0000
5.3 26250 0 0 0 8932714 0
0 10086 active+clean 20h 8093'56786
8093:5109316 [0,26,32]p0 [0,26,32]p0 2023-07-
31T11:02:35.864452+0000 2023-07-30T04:25:56.495524+0000
5 periodic scrub scheduled @ 2023-08-01T20:18:30.975924+0000
5.4 26071 0 0 0 8874237 0
0 10024 active+clean 6h 8092'53824
8093:5146409 [15,7,34]p15 [15,7,34]p15 2023-08-
01T01:16:48.361184+0000 2023-07-25T15:47:10.627594+0000
5 periodic scrub scheduled @ 2023-08-02T12:32:06.359395+0000
5.5 26160 0 0 0 8870317 0
0 10073 active+clean 12h 8093'56173
8093:4706658 [9,31,16]p9 [9,31,16]p9 2023-07-
31T18:52:26.301525+0000 2023-07-29T08:19:00.537322+0000
5 periodic scrub scheduled @ 2023-08-02T02:11:16.267794+0000
5.6 26186 0 0 0 8904446 0
0 10084 active+clean 44m 8093'57584
8093:5032349 [7,10,38]p7 [7,10,38]p7 2023-08-
01T06:37:45.184419+0000 2023-08-01T06:37:45.184419+0000
313 periodic scrub scheduled @ 2023-08-02T10:01:32.285716+0000
5.7 26292 0 0 0 8908213 0
0 9695 active+clean 87m 8093'56896
8093:4969718 [36,1,13]p36 [36,1,13]p36 2023-08-
01T05:55:06.016287+0000 2023-07-30T21:49:33.028594+0000
5 periodic scrub scheduled @ 2023-08-02T13:54:38.778542+0000
5.8 26323 0 0 0 8911110 0
0 9747 active+clean 3h 8093'56448
8093:4981465 [36,15,2]p36 [36,15,2]p36 2023-08-
01T04:21:06.360778+0000 2023-07-29T14:46:02.363530+0000
5 periodic scrub scheduled @ 2023-08-02T10:36:48.764085+0000
5.9 26035 0 0 0 8829430 0
0 10034 active+clean 20h 8093'56335
8093:4829155 [37,21,24]p37 [37,21,24]p37 2023-07-
31T11:07:39.961751+0000 2023-07-31T11:07:39.961751+0000
309 periodic scrub scheduled @ 2023-08-01T22:42:18.862879+0000
5.a 26052 0 0 0 8859067 0
0 10087 active+clean 27h 8092'56087
8093:5022933 [2,23,10]p2 [2,23,10]p2 2023-07-
31T03:28:44.433360+0000 2023-07-31T03:28:44.433360+0000
248 periodic scrub scheduled @ 2023-08-01T13:36:39.897693+0000
5.b 25759 0 0 0 8739834 0
0 9693 active+clean 15h 8090'56293
8093:4837010 [36,7,28]p36 [36,7,28]p36 2023-07-
31T15:55:00.415967+0000 2023-07-31T15:55:00.415967+0000
323 periodic scrub scheduled @ 2023-08-01T23:41:03.756058+0000
5.c 25927 0 0 0 8788271 78051
398 10051 active+clean 24h 8093'174851
8093:4982667 [5,36,18]p5 [5,36,18]p5 2023-07-
31T07:20:32.208533+0000 2023-07-31T07:20:32.208533+0000
315 periodic scrub scheduled @ 2023-08-01T18:18:15.292651+0000
5.d 25995 0 0 0 8815306 0
0 10070 active+clean 2h 8093'57270
8093:4994478 [32,13,16]p32 [32,13,16]p32 2023-08-
01T04:27:55.863933+0000 2023-08-01T04:27:55.863933+0000
294 periodic scrub scheduled @ 2023-08-02T05:55:30.108279+0000
5.e 26253 0 0 0 8939984 0
0 10018 active+clean 5h 8092'56919
8093:5135033 [37,19,4]p37 [37,19,4]p37 2023-08-
01T01:38:15.740983+0000 2023-07-30T21:55:45.349878+0000
5 periodic scrub scheduled @ 2023-08-02T04:40:09.172157+0000
5.f 26025 0 0 0 8821973 0
0 10020 active+clean 25h 8093'53120
8093:4794909 [14,11,8]p14 [14,11,8]p14 2023-07-
31T06:22:22.849194+0000 2023-07-25T02:42:24.135997+0000
5 periodic deep scrub scheduled @ 2023-08-01T13:50:38.914828+0000
5.10 25999 0 0 0 8821303 0
0 10048 active+clean 3h 8092'55848
8093:5151525 [39,27,14]p39 [39,27,14]p39 2023-08-
01T03:40:44.355521+0000 2023-07-29T23:58:49.567904+0000
5 periodic scrub scheduled @ 2023-08-02T04:44:52.459615+0000
5.11 26200 0 0 0 8897148 0
0 8909 active+clean 23h 8093'56309
8093:4858657 [35,24,23]p35 [35,24,23]p35 2023-07-
31T08:19:21.090885+0000 2023-07-30T07:21:21.135342+0000
5 periodic scrub scheduled @ 2023-08-01T13:09:38.620801+0000
5.12 26043 0 0 0 8803496 0
0 10016 active+clean 33h 8093'52716
8093:5090415 [21,35,3]p21 [21,35,3]p21 2023-07-
30T22:10:31.308788+0000 2023-07-24T14:40:34.453392+0000
5 periodic deep scrub scheduled @ 2023-08-01T09:36:02.058413+0000
5.13 25929 0 0 0 8785411 0
0 10029 active+clean 16h 8090'54629
8093:5096641 [32,17,9]p32 [32,17,9]p32 2023-07-
31T15:19:04.119491+0000 2023-07-27T16:36:53.401620+0000
5 periodic scrub scheduled @ 2023-08-01T22:19:03.573063+0000
5.14 26102 0 0 0 8858274 0
0 10069 active+clean 89m 8093'54671
8093:4958083 [3,29,12]p3 [3,29,12]p3 2023-08-
01T05:53:20.722831+0000 2023-07-27T11:29:26.930179+0000
5 periodic scrub scheduled @ 2023-08-02T14:03:11.521746+0000
5.15 26122 0 0 0 8850708 0
0 9419 active+clean 14h 8093'57119
8093:4854254 [28,8,29]p28 [28,8,29]p28 2023-07-
31T17:04:04.790500+0000 2023-07-31T17:04:04.790500+0000
309 periodic scrub scheduled @ 2023-08-02T04:29:01.728903+0000
5.16 26168 0 0 0 8869093 80186950
401505 10031 active+clean 32h 8093'122396977
8093:127435157 [26,39,9]p26 [26,39,9]p26 2023-07-
30T23:19:12.784044+0000 2023-07-30T23:19:12.784044+0000
258 periodic scrub scheduled @ 2023-08-01T10:29:11.462563+0000
5.17 25504 0 0 0 8634818 0
0 10081 active+clean 23h 8093'55481
8093:4742014 [4,25,34]p4 [4,25,34]p4 2023-07-
31T08:11:23.105601+0000 2023-07-31T08:11:23.105601+0000
309 periodic scrub scheduled @ 2023-08-01T15:11:19.240302+0000
5.18 26143 0 0 0 8846680 0
0 10014 active+clean 23h 8093'55015
8093:4927120 [22,11,36]p22 [22,11,36]p22 2023-07-
31T08:18:27.757381+0000 2023-07-27T19:04:02.036522+0000
5 periodic scrub scheduled @ 2023-08-01T16:25:12.741673+0000
5.19 26117 0 0 0 8864860 0
0 10073 active+clean 27h 8093'55173
8093:5001362 [1,28,27]p1 [1,28,27]p1 2023-07-
31T03:37:40.525594+0000 2023-07-28T10:50:18.232627+0000
5 periodic scrub scheduled @ 2023-08-01T15:06:48.300456+0000
5.1a 26186 0 0 0 8870466 0
0 10025 active+clean 31h 8093'54025
8093:4991279 [34,17,22]p34 [34,17,22]p34 2023-07-
31T00:20:28.853158+0000 2023-07-25T21:31:55.045662+0000
5 periodic scrub scheduled @ 2023-08-01T09:42:20.646380+0000
5.1b 26070 0 0 0 8854703 0
0 10087 active+clean 8h 8093'56487
8093:4886996 [22,7,37]p22 [22,7,37]p22 2023-07-
31T22:39:17.793412+0000 2023-07-30T16:07:40.211725+0000
6 periodic scrub scheduled @ 2023-08-01T23:23:58.360930+0000
5.1c 26302 0 0 0 8925675 0
0 10015 active+clean 16h 8093'54915
8093:4986627 [33,11,2]p33 [33,11,2]p33 2023-07-
31T15:07:27.474683+0000 2023-07-27T15:11:42.794360+0000
5 periodic scrub scheduled @ 2023-08-01T15:51:01.248015+0000
5.1d 26075 0 0 0 8839499 0
0 10075 active+clean 21h 8093'52575
8093:4857107 [33,16,17]p33 [33,16,17]p33 2023-07-
31T09:23:24.577919+0000 2023-07-24T16:52:37.965968+0000
5 periodic deep scrub scheduled @ 2023-08-01T11:33:35.067584+0000
5.1e 26182 0 0 0 8905099 0
0 8905 active+clean+scrubbing+deep 4m 8093'56905
8093:4936987 [35,20,15]p35 [35,20,15]p35 2023-07-
31T04:25:51.171690+0000 2023-07-24T14:28:19.531701+0000
5 deep scrubbing for 259s
5.1f 25978 0 0 0 8796478 0
0 10068 active+clean 13h 8092'56868
8093:4813791 [26,30,13]p26 [26,30,13]p26 2023-07-
31T17:50:40.349450+0000 2023-07-31T17:50:40.349450+0000
311 periodic scrub scheduled @ 2023-08-02T04:39:41.913504+0000
On Tue, 2023-08-01 at 06:14 +0000, Eugen Block wrote:
Yeah, regarding data distribution increasing the pg_num of the data
pool is recommended. But could you also share the output of:
ceph pg ls-by-pool default.rgw.meta
That's where the large omap was reported, maybe you'll need to
increase the pg_num for that pool as well. Personally, I always
disable the autoscaler.
Zitat von Mark Johnson
<markj@xxxxxxxxx<mailto:markj@xxxxxxxxx><mailto:markj@xxxxxxxxx<mailt
o:markj@xxxxxxxxx>>>:
Thanks Bailey,
With regards to the PG count, we've been relying on PG autoscale
and
it is currently enabled. I figure I'd need to disable autoscale
and
manually increase the PG count on the default.rgw.buckets.data
pool,
correct? We're coming from our existing clusters running Jewel to
this new Quincy cluster and have no prior experience with
autoscale
so we were just assuming autoscale would manage PG counts better
than us doing it manually. As you can probably guess, we don't
have
much experience with Ceph.
Regards,
Mark Johnson
On Mon, 2023-07-31 at 21:54 -0300, Bailey Allison wrote:
[You don't often get email from
ballison@xxxxxxxxxxxx<mailto:ballison@xxxxxxxxxxxx><mailto:ballison@4
5drives.com<mailto:ballison@xxxxxxxxxxxx>><mailto:ballison@45drives.c
om<mailto:ballison@xxxxxxxxxxxx><mailto:ballison@xxxxxxxxxxxx<mailto:
ballison@xxxxxxxxxxxx>>>. Learn why
this
is important at https://aka.ms/LearnAboutSenderIdentification ;]
Hi,
It appears you have quite a low PG count on your cluster (approx.
20
PGs per each OSD).
Usually is recommended to have about 100-150 per each OSD. With a
lower PG count you can have issues with balancing data and cause
errors such as large OMAP objects.
Might not be the fix in this case but either way would still
recommend increasing PGs on your pools.
If you look at the OMAP value in your ceph osd df you can see that
some OSDs have 2GB while some have 500MB. Even for data some
drives
are holding 900GB while others 2TB.
You will have to issue a deep-scrub on the PGs as well to get
updated OMAP data once the PGs are increased.
Regards,
Bailey
-----Original Message-----
From: Mark Johnson
<markj@xxxxxxxxx<mailto:markj@xxxxxxxxx><mailto:markj@xxxxxxxxx<mailt
o:markj@xxxxxxxxx>><mailto:markj@xxxxxxxxx<mailto:markj@xxxxxxxxx><ma
ilto:markj@xxxxxxxxx<mailto:markj@xxxxxxxxx>>>>
Sent: July 31, 2023 9:01 PM
To:
eblock@xxxxxx<mailto:eblock@xxxxxx><mailto:eblock@xxxxxx<mailto:ebloc
k@xxxxxx>><mailto:eblock@xxxxxx<mailto:eblock@xxxxxx><mailto:eblock@n
de.ag<mailto:eblock@xxxxxx>>>;
ceph-users@xxxxxxx<mailto:ceph-users@xxxxxxx><mailto:ceph-
users@xxxxxxx<mailto:ceph-users@xxxxxxx>><mailto:ceph-
users@xxxxxxx<mailto:ceph-users@xxxxxxx><mailto:ceph-
users@xxxxxxx<mailto:ceph-users@xxxxxxx>>>
Subject: Re: 1 Large omap object found
Sure thing. Thanks for the reply.
ceph df
--- RAW STORAGE ---
CLASS SIZE AVAIL USED RAW USED %RAW USED
hdd 291 TiB 244 TiB 47 TiB 47 TiB 16.02
TOTAL 291 TiB 244 TiB 47 TiB 47 TiB 16.02
--- POOLS ---
POOL ID PGS STORED OBJECTS USED
%USED
MAX AVAIL
.mgr 1 1 459 MiB 116 1.3 GiB
65 TiB
.rgw.root 2 32 1.3 KiB 4 48 KiB
65 TiB
default.rgw.log 3 32 5.3 KiB 209 468 KiB
65 TiB
default.rgw.control 4 32 0 B 8 0 B
65 TiB
default.rgw.meta 5 32 452 MiB 828.75k 10 GiB
65 TiB
default.rgw.buckets.index 6 32 17 GiB 4.56M 51 GiB
0.03
65 TiB
default.rgw.buckets.data 7 128 15 TiB 54.51M 46 TiB
19.24
65 TiB
cephfs_metadata 8 16 258 MiB 98 775 MiB
65 TiB
cephfs_data 9 32 1.9 GiB 998 5.6 GiB
65 TiB
ceph osd df
ID CLASS WEIGHT REWEIGHT SIZE RAW USE DATA OMAP
META
AVAIL %USE VAR PGS STATUS
0 hdd 7.27739 1.00000 7.3 TiB 1.6 TiB 1.6 TiB 550 MiB
12
GiB 5.7 TiB 21.70 1.35 21 up
1 hdd 7.27739 1.00000 7.3 TiB 995 GiB 986 GiB 1.1 GiB
7.6
GiB 6.3 TiB 13.35 0.83 28 up
2 hdd 7.27739 1.00000 7.3 TiB 996 GiB 986 GiB 2.1 GiB
7.9
GiB 6.3 TiB 13.37 0.83 22 up
3 hdd 7.27739 1.00000 7.3 TiB 1.3 TiB 1.3 TiB 513 MiB
10
GiB 5.9 TiB 18.35 1.15 28 up
4 hdd 7.27739 1.00000 7.3 TiB 1.1 TiB 1.1 TiB 527 MiB
8.3
GiB 6.2 TiB 15.02 0.94 22 up
5 hdd 7.27739 1.00000 7.3 TiB 1.8 TiB 1.8 TiB 1.5 GiB
14
GiB 5.5 TiB 25.01 1.56 28 up
6 hdd 7.27739 1.00000 7.3 TiB 746 GiB 739 GiB 1.0 GiB
5.8
GiB 6.5 TiB 10.01 0.63 20 up
7 hdd 7.27739 1.00000 7.3 TiB 1.1 TiB 1.1 TiB 1.1 GiB
8.8
GiB 6.2 TiB 15.04 0.94 20 up
8 hdd 7.27739 1.00000 7.3 TiB 871 GiB 864 GiB 544 MiB
6.7
GiB 6.4 TiB 11.69 0.73 27 up
9 hdd 7.27739 1.00000 7.3 TiB 1.3 TiB 1.3 TiB 3.3 GiB
11
GiB 5.9 TiB 18.37 1.15 28 up
30 hdd 7.27739 1.00000 7.3 TiB 1.8 TiB 1.8 TiB 1.6 GiB
14
GiB 5.5 TiB 25.01 1.56 35 up
31 hdd 7.27739 1.00000 7.3 TiB 747 GiB 739 GiB 2.2 GiB
6.2
GiB 6.5 TiB 10.03 0.63 20 up
32 hdd 7.27739 1.00000 7.3 TiB 996 GiB 987 GiB 1.5 GiB
7.9
GiB 6.3 TiB 13.37 0.83 26 up
33 hdd 7.27739 1.00000 7.3 TiB 995 GiB 985 GiB 1.5 GiB
7.7
GiB 6.3 TiB 13.35 0.83 25 up
34 hdd 7.27739 1.00000 7.3 TiB 750 GiB 742 GiB 2.1 GiB
5.7
GiB 6.5 TiB 10.07 0.63 25 up
35 hdd 7.27739 1.00000 7.3 TiB 2.1 TiB 2.0 TiB 571 MiB
15
GiB 5.2 TiB 28.36 1.77 34 up
36 hdd 7.27739 1.00000 7.3 TiB 1.3 TiB 1.3 TiB 1.5 GiB
10
GiB 5.9 TiB 18.37 1.15 31 up
37 hdd 7.27739 1.00000 7.3 TiB 1.1 TiB 1.1 TiB 524 MiB
8.2
GiB 6.2 TiB 14.99 0.94 26 up
38 hdd 7.27739 1.00000 7.3 TiB 1.6 TiB 1.6 TiB 1.6 GiB
12
GiB 5.7 TiB 21.70 1.35 28 up
39 hdd 7.27739 1.00000 7.3 TiB 1.5 TiB 1.4 TiB 2.4 GiB
11
GiB 5.8 TiB 20.04 1.25 30 up
10 hdd 7.27739 1.00000 7.3 TiB 1.3 TiB 1.3 TiB 1.6 GiB
10
GiB 5.9 TiB 18.34 1.14 26 up
12 hdd 7.27739 1.00000 7.3 TiB 1.3 TiB 1.3 TiB 1 KiB
9.9
GiB 5.9 TiB 18.37 1.15 25 up
14 hdd 7.27739 1.00000 7.3 TiB 1.5 TiB 1.4 TiB 593 MiB
10
GiB 5.8 TiB 19.98 1.25 22 up
16 hdd 7.27739 1.00000 7.3 TiB 997 GiB 987 GiB 2.2 GiB
7.5
GiB 6.3 TiB 13.38 0.84 19 up
18 hdd 7.27739 1.00000 7.3 TiB 1.1 TiB 1.1 TiB 1.1 GiB
8.7
GiB 6.2 TiB 15.02 0.94 26 up
20 hdd 7.27739 1.00000 7.3 TiB 1.6 TiB 1.6 TiB 1.1 GiB
12
GiB 5.7 TiB 21.68 1.35 26 up
22 hdd 7.27739 1.00000 7.3 TiB 1.3 TiB 1.3 TiB 559 MiB
10
GiB 5.9 TiB 18.34 1.14 22 up
24 hdd 7.27739 1.00000 7.3 TiB 872 GiB 864 GiB 1020 MiB
6.8
GiB 6.4 TiB 11.70 0.73 23 up
26 hdd 7.27739 1.00000 7.3 TiB 749 GiB 741 GiB 1.8 GiB
6.3
GiB 6.5 TiB 10.05 0.63 25 up
28 hdd 7.27739 1.00000 7.3 TiB 1.3 TiB 1.3 TiB 1.5 GiB
10
GiB 5.9 TiB 18.36 1.15 32 up
11 hdd 7.27739 1.00000 7.3 TiB 1.1 TiB 1.1 TiB 2.6 GiB
8.5
GiB 6.2 TiB 15.02 0.94 23 up
13 hdd 7.27739 1.00000 7.3 TiB 1.3 TiB 1.3 TiB 2.2 GiB
10
GiB 5.9 TiB 18.38 1.15 36 up
15 hdd 7.27739 1.00000 7.3 TiB 995 GiB 986 GiB 1.1 GiB
7.7
GiB 6.3 TiB 13.35 0.83 25 up
17 hdd 7.27739 1.00000 7.3 TiB 623 GiB 618 GiB 419 KiB
5.0
GiB 6.7 TiB 8.35 0.52 23 up
19 hdd 7.27739 1.00000 7.3 TiB 870 GiB 863 GiB 513 MiB
6.6
GiB 6.4 TiB 11.67 0.73 21 up
21 hdd 7.27739 1.00000 7.3 TiB 1.1 TiB 1.1 TiB 1.5 GiB
8.6
GiB 6.2 TiB 15.02 0.94 25 up
23 hdd 7.27739 1.00000 7.3 TiB 746 GiB 739 GiB 564 MiB
5.8
GiB 6.5 TiB 10.01 0.62 22 up
25 hdd 7.27739 1.00000 7.3 TiB 1.1 TiB 1.1 TiB 2.1 GiB
8.4
GiB 6.2 TiB 15.03 0.94 24 up
27 hdd 7.27739 1.00000 7.3 TiB 1.2 TiB 1.2 TiB 532 MiB
9.1
GiB 6.1 TiB 16.68 1.04 23 up
29 hdd 7.27739 1.00000 7.3 TiB 1.1 TiB 1.1 TiB 1.1 GiB
8.4
GiB 6.2 TiB 14.99 0.94 19 up
TOTAL 291 TiB 47 TiB 46 TiB 51 GiB
359
GiB 244 TiB 16.02
MIN/MAX VAR: 0.52/1.77 STDDEV: 4.56
On Mon, 2023-07-31 at 09:22 +0000, Eugen Block wrote:
Hi,
can you share some more details like 'ceph df' and 'ceph osd df'? I
don't have too much advice yet, but to see all entries in your meta
pool you need add the --all flag because those objects are stored
in
namespaces:
rados -p default.rgw.meta ls --all
That pool contains user and bucket information (example):
# rados -p default.rgw.meta ls --all
users.uid admin.buckets
users.keys c0fba3ea7d9c4321b5205752c85baa85 users.uid
admin
users.keys JBWPRAPP1AQG471AMGC4 users.uid
e434b82737cf4138b899c0785b49112d.buckets
users.uid e434b82737cf4138b899c0785b49112d
Zitat von Mark Johnson
<markj@xxxxxxxxx<mailto:markj@xxxxxxxxx><mailto:markj@xxxxxxxxx<mailt
o:markj@xxxxxxxxx>><mailto:markj@xxxxxxxxx<mailto:markj@xxxxxxxxx><ma
ilto:markj@xxxxxxxxx<mailto:markj@xxxxxxxxx>>>>:
I've been going round and round in circles trying to work this one
out but I'm getting nowhere. We're running a 4 node quincy cluster
(17.2.6) which recently reported the following:
ceph.log-20230729.gz:2023-07-28T08:31:42.390003+0000 osd.26
(osd.26)
13834 : cluster [WRN] Large omap object found. Object:
5:6c65dd84:users.uid::callrecordings$callrecordings_rw.buckets:head
PG: 5.21bba636 (5.16) Key count: 378454 Size (bytes): 75565579
This happened a week or so ago (only the key count was only just
over the 200000 threshold on that occasion) and after much
searching
around, I found an article that suggested a deep scrub on the pg
would likely resolve the issue, so I forced a deep scrub and
shortly
after, the warning cleared. Came into the office today to discover
the above. It's on the same PG as before which is in the
default.rgw.meta pool. This time, after forcing a deep-scrub on
that PG, nothing changed. I did it a second time just to be sure
but got the same result.
I keep finding a suse article that simply suggests increasing the
threshold to the previous default of 2,000,000, but other articles
I
read say it was lowered for a reason and that by the time it hits
that figure, it's too late so I don't want to just mask it.
Problem
is that I don't really understand it. I found a thread here from
a
bit over two years ago but their issue was in the
default.rgw.buckets.index pool. A step in the solution was to list
out the problematic object id and check the objects per shard
however, if I issue the command "rados -p default.rgw.meta ls" it
returns nothing. I get a big list from "rados -p
default.rgw.buckets.index ls" just nothing from the first pool. I
think it may be because the meta pool isn't indexed based on
something I read, but I really don't know what I'm talking about
tbh.
I don't know if this is helpful, but if I list out all the PGs for
that pool, there are 32 PGs and 5.16 shows 80186950 bytes and
401505
keys. PG 5.c has 75298 and 384 keys. The remaining 30 PGs show
zero bytes and zero keys. I'm really not sure how to troubleshoot
and resolve from here. For the record, dynamic resharding is
enabled in that no options have been set in the config and that is
the default setting.
Based on the suse article I mentioned which also references the
default.rgw.meta pool, I'm gathering our issue is because we have
so
many buckets that are all owned by the one user and the solution is
either:
* delete unused buckets
* create multiple users and spread buckets evenly across all users
(not something we can do)
* increase the threshold to stop the warning
Problem is that I'm having trouble verifying this is the issue.
I've
tried dumping out bucket stats to a file (radosgw-admin bucket
stats
bucket_stats.txt) but after three hours this is still running with
no output.
Thanks for your time,
Mark
_______________________________________________
ceph-users mailing list --
ceph-users@xxxxxxx<mailto:ceph-users@xxxxxxx><mailto:ceph-
users@xxxxxxx<mailto:ceph-users@xxxxxxx>><mailto:ceph-
users@xxxxxxx<mailto:ceph-users@xxxxxxx><mailto:ceph-
users@xxxxxxx<mailto:ceph-users@xxxxxxx>>> To unsubscribe send
an
email to ceph-users-leave@xxxxxxx<mailto:ceph-users-
leave@xxxxxxx><mailto:ceph-users-leave@xxxxxxx<mailto:ceph-users-
leave@xxxxxxx>><mailto:ceph-users-leave@xxxxxxx<mailto:ceph-users-
leave@xxxxxxx><mailto:ceph-users-leave@xxxxxxx<mailto:ceph-users-
leave@xxxxxxx>>>
_______________________________________________
ceph-users mailing list --
ceph-users@xxxxxxx<mailto:ceph-users@xxxxxxx><mailto:ceph-
users@xxxxxxx<mailto:ceph-users@xxxxxxx>><mailto:ceph-
users@xxxxxxx<mailto:ceph-users@xxxxxxx><mailto:ceph-
users@xxxxxxx<mailto:ceph-users@xxxxxxx>>> To unsubscribe send
an
email to ceph-users-leave@xxxxxxx<mailto:ceph-users-
leave@xxxxxxx><mailto:ceph-users-leave@xxxxxxx<mailto:ceph-users-
leave@xxxxxxx>><mailto:ceph-users-leave@xxxxxxx<mailto:ceph-users-
leave@xxxxxxx><mailto:ceph-users-leave@xxxxxxx<mailto:ceph-users-
leave@xxxxxxx>>>
_______________________________________________
ceph-users mailing list --
ceph-users@xxxxxxx<mailto:ceph-users@xxxxxxx><mailto:ceph-
users@xxxxxxx<mailto:ceph-users@xxxxxxx>><mailto:ceph-
users@xxxxxxx<mailto:ceph-users@xxxxxxx><mailto:ceph-
users@xxxxxxx<mailto:ceph-users@xxxxxxx>>> To unsubscribe send
an
email to ceph-users-leave@xxxxxxx<mailto:ceph-users-
leave@xxxxxxx><mailto:ceph-users-leave@xxxxxxx<mailto:ceph-users-
leave@xxxxxxx>><mailto:ceph-users-leave@xxxxxxx<mailto:ceph-users-
leave@xxxxxxx><mailto:ceph-users-leave@xxxxxxx<mailto:ceph-users-
leave@xxxxxxx>>>
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx<mailto:ceph-
users@xxxxxxx>
To unsubscribe send an email to ceph-users-leave@xxxxxxx<mailto:ceph-
users-leave@xxxxxxx>
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx
