I installed a rook.io ceph storage cluster. Before installation, I cleaned up the previous installation like described here: https://rook.io/docs/rook/v1.7/ceph-teardown.html
The new cluster was provisioned correctly, however ceph is not healthy immediately after provisioning, and stuck.
data:
pools: 1 pools, 128 pgs
objects: 0 objects, 0 B
usage: 20 MiB used, 15 TiB / 15 TiB avail
pgs: 100.000% pgs not active
128 undersized+peered
[root#rook-ceph-tools-74df559676-scmzg /]# ceph osd df
ID CLASS WEIGHT REWEIGHT SIZE RAW USE DATA OMAP META AVAIL %USE VAR PGS STATUS
0 hdd 3.63869 1.00000 3.6 TiB 5.0 MiB 144 KiB 0 B 4.8 MiB 3.6 TiB 0 0.98 0 up
1 hdd 3.63869 1.00000 3.6 TiB 5.4 MiB 144 KiB 0 B 5.2 MiB 3.6 TiB 0 1.07 128 up
2 hdd 3.63869 1.00000 3.6 TiB 5.0 MiB 144 KiB 0 B 4.8 MiB 3.6 TiB 0 0.98 0 up
3 hdd 3.63869 1.00000 3.6 TiB 4.9 MiB 144 KiB 0 B 4.8 MiB 3.6 TiB 0 0.97 0 up
TOTAL 15 TiB 20 MiB 576 KiB 0 B 20 MiB 15 TiB 0
MIN/MAX VAR: 0.97/1.07 STDDEV: 0
[root#rook-ceph-tools-74df559676-scmzg /]# ceph osd tree
ID CLASS WEIGHT TYPE NAME STATUS REWEIGHT PRI-AFF
-1 14.55475 root default
-3 14.55475 host storage1-kube-domain-tld
0 hdd 3.63869 osd.0 up 1.00000 1.00000
1 hdd 3.63869 osd.1 up 1.00000 1.00000
2 hdd 3.63869 osd.2 up 1.00000 1.00000
3 hdd 3.63869 osd.3 up 1.00000 1.00000
Is there anyone who can explain what went wrong and how to fix the issue?
The problem is that osds are running on the same host and failure domain is set to host. Switching failure domain to osd fixes the issue. The default failure domain can be changed as per https://stackoverflow.com/a/63472905/3146709
Related
Ceph cluster shows following weird behavior with ceph df output:
--- RAW STORAGE ---
CLASS SIZE AVAIL USED RAW USED %RAW USED
hdd 817 TiB 399 TiB 418 TiB 418 TiB 51.21
ssd 1.4 TiB 1.2 TiB 22 GiB 174 GiB 12.17
TOTAL 818 TiB 400 TiB 418 TiB 419 TiB 51.15
--- POOLS ---
POOL ID PGS STORED OBJECTS USED %USED MAX AVAIL
pool1 45 300 21 TiB 6.95M 65 TiB 20.23 85 TiB
pool2 50 50 72 GiB 289.15k 357 GiB 0.14 85 TiB
pool3 53 64 2.9 TiB 754.06k 8.6 TiB 3.24 85 TiB
erasurepool_data 57 1024 138 TiB 50.81M 241 TiB 48.49 154 TiB
erasurepool_metadata 58 8 9.1 GiB 1.68M 27 GiB 2.46 362 GiB
device_health_metrics 59 1 22 MiB 163 66 MiB 0 85 TiB
.rgw.root 60 8 5.6 KiB 17 3.5 MiB 0 85 TiB
.rgw.log 61 8 70 MiB 2.56k 254 MiB 0 85 TiB
.rgw.control 62 8 0 B 8 0 B 0 85 TiB
.rgw.meta 63 8 7.6 MiB 52 32 MiB 0 85 TiB
.rgw.buckets.index 64 8 11 GiB 1.69k 34 GiB 3.01 362 GiB
.rgw.buckets.data 65 512 23 TiB 33.87M 72 TiB 21.94 85 TiB
As seen above available storage 399TiB, and max avail in pool list shows 85TiB. I use 3 replicas for each pool replicated pool and 3+2 erasure code for the erasurepool_data.
As far as I know Max Avail segment shows max raw available capacity according to replica size. So it comes up to 85*3=255TiB. Meanwhile cluster shows almost 400 available.
Which to trust?
Is this only a bug?
Turns out max available space is calculated according to the fullest osds in the cluster and has nothing to do with total free space in the cluster. From what i've found this kind of fluctiation mainly happens on small clusters.
MAX AVAIL column represents the amount of data that can be used before the first OSD becomes full. It takes into account the projected distribution of data across disks from the CRUSH map and uses the 'first OSD to fill up' as the target. it does not seem to be a bug. If MAX AVAIL is not what you expect it to be, look at the data distribution using ceph osd tree and make sure you have a uniform distribution.
You can also check some helpful posts here that explains some of the miscalculations:
Using available space in a Ceph pool
ceph-displayed-size-calculation
max-avail-in-ceph-df-command-is-incorrec
As you have Erasure Coding involved please check this SO post:
ceph-df-octopus-shows-used-is-7-times-higher-than-stored-in-erasure-coded-pool
When you add the erasure coded pool, i.e. erasurepool_data at 154, you get 255+154 = 399.
The pool default.rgw.buckets.data has 501 GiB stored, but USED shows 3.5 TiB.
root#ceph-01:~# ceph df
--- RAW STORAGE ---
CLASS SIZE AVAIL USED RAW USED %RAW USED
hdd 196 TiB 193 TiB 3.5 TiB 3.6 TiB 1.85
TOTAL 196 TiB 193 TiB 3.5 TiB 3.6 TiB 1.85
--- POOLS ---
POOL ID PGS STORED OBJECTS USED %USED MAX AVAIL
device_health_metrics 1 1 19 KiB 12 56 KiB 0 61 TiB
.rgw.root 2 32 2.6 KiB 6 1.1 MiB 0 61 TiB
default.rgw.log 3 32 168 KiB 210 13 MiB 0 61 TiB
default.rgw.control 4 32 0 B 8 0 B 0 61 TiB
default.rgw.meta 5 8 4.8 KiB 11 1.9 MiB 0 61 TiB
default.rgw.buckets.index 6 8 1.6 GiB 211 4.7 GiB 0 61 TiB
default.rgw.buckets.data 10 128 501 GiB 5.36M 3.5 TiB 1.90 110 TiB
The default.rgw.buckets.data pool is using erasure coding:
root#ceph-01:~# ceph osd erasure-code-profile get EC_RGW_HOST
crush-device-class=hdd
crush-failure-domain=host
crush-root=default
jerasure-per-chunk-alignment=false
k=6
m=4
plugin=jerasure
technique=reed_sol_van
w=8
If anyone could help explain why it's using up 7 times more space, it would help a lot.
Versioning is disabled. ceph version 15.2.13 (octopus stable).
This is related to bluestore_min_alloc_size_hdd=64K (default on Octopus).
If using Erasure Coding, data is broken up into smaller chunks, which each take 64K on disk.
One option would be to lower bluestore_min_alloc_size_hdd to 4K, which is good if your solution requires storing millions of tiny (16K) objects. In my case, I'm storing hundreds of millions of 3-4M photos, so I decide to skip Erasure Coding, stay on bluestore_min_alloc_size_hdd=64K and switch to replicated 3 (min 2). Which is much safer and faster in the long run.
Here is the reply from Josh Baergen on the mailing list:
Hey Arkadiy,
If the OSDs are on HDDs and were created with the default
bluestore_min_alloc_size_hdd, which is still 64KiB in Octopus, then in
effect data will be allocated from the pool in 640KiB chunks (64KiB *
(k+m)). 5.36M objects taking up 501GiB is an average object size of 98KiB
which results in a ratio of 6.53:1 allocated:stored, which is pretty close
to the 7:1 observed.
If my assumption about your configuration is correct, then the only way to
fix this is to adjust bluestore_min_alloc_size_hdd and recreate all your
OSDs, which will take a while...
Josh
First of all sorry for my poor English
In my ceph cluster, when i run the ceph df detail command it shows me like as following result
RAW STORAGE:
CLASS SIZE AVAIL USED RAW USED %RAW USED
hdd 62 TiB 52 TiB 10 TiB 10 TiB 16.47
ssd 8.7 TiB 8.4 TiB 370 GiB 377 GiB 4.22
TOTAL 71 TiB 60 TiB 11 TiB 11 TiB 14.96
POOLS:
POOL ID STORED OBJECTS USED %USED MAX AVAIL QUOTA OBJECTS QUOTA BYTES DIRTY USED COMPR UNDER COMPR
rbd-kubernetes 36 288 GiB 71.56k 865 GiB 1.73 16 TiB N/A N/A 71.56k 0 B 0 B
rbd-cache 41 2.4 GiB 208.09k 7.2 GiB 0.09 2.6 TiB N/A N/A 205.39k 0 B 0 B
cephfs-metadata 51 529 MiB 221 1.6 GiB 0 16 TiB N/A N/A 221 0 B 0 B
cephfs-data 52 1.0 GiB 424 3.1 GiB 0 16 TiB N/A N/A 424 0 B 0 B
So i have a question about the result
As you can see, sum of my pools used storage is less than 1 TB, But in RAW STORAGE section the used from HDD hard disks is 10TB and it is growing every day.I think this is unusual and something is wrong with this CEPH cluster.
And also FYI the output of ceph osd dump | grep replicated is
pool 36 'rbd-kubernetes' replicated size 3 min_size 2 crush_rule 0 object_hash rjenkins pg_num 256 pgp_num 244 pg_num_target 64 pgp_num_target 64 last_change 1376476 lfor 2193/2193/2193 flags hashpspool,selfmanaged_snaps,creating tiers 41 read_tier 41 write_tier 41 stripe_width 0 application rbd
pool 41 'rbd-cache' replicated size 3 min_size 2 crush_rule 1 object_hash rjenkins pg_num 64 pgp_num 64 autoscale_mode on last_change 1376476 lfor 2193/2193/2193 flags hashpspool,incomplete_clones,selfmanaged_snaps,creating tier_of 36 cache_mode writeback target_bytes 1000000000000 hit_set bloom{false_positive_probability: 0.05, target_size: 0, seed: 0} 3600s x1 decay_rate 0 search_last_n 0 min_read_recency_for_promote 1 min_write_recency_for_promote 1 stripe_width 0
pool 51 'cephfs-metadata' replicated size 3 min_size 2 crush_rule 0 object_hash rjenkins pg_num 32 pgp_num 32 autoscale_mode on last_change 31675 flags hashpspool stripe_width 0 pg_autoscale_bias 4 pg_num_min 16 recovery_priority 5 application cephfs
pool 52 'cephfs-data' replicated size 3 min_size 2 crush_rule 0 object_hash rjenkins pg_num 32 pgp_num 32 autoscale_mode on last_change 742334 flags hashpspool,selfmanaged_snaps stripe_width 0 application cephfs
Ceph Version ceph -v
ceph version 14.2.10 (b340acf629a010a74d90da5782a2c5fe0b54ac20) nautilus (stable)
Ceph OSD versions ceph tell osd.* version return for all OSDs like
osd.0: {
"version": "ceph version 14.2.10 (b340acf629a010a74d90da5782a2c5fe0b54ac20) nautilus (stable)"
}
Ceph status ceph -s
cluster:
id: 6a86aee0-3171-4824-98f3-2b5761b09feb
health: HEALTH_OK
services:
mon: 3 daemons, quorum ceph-sn-03,ceph-sn-02,ceph-sn-01 (age 37h)
mgr: ceph-sn-01(active, since 4d), standbys: ceph-sn-03, ceph-sn-02
mds: cephfs-shared:1 {0=ceph-sn-02=up:active} 2 up:standby
osd: 63 osds: 63 up (since 41h), 63 in (since 41h)
task status:
scrub status:
mds.ceph-sn-02: idle
data:
pools: 4 pools, 384 pgs
objects: 280.29k objects, 293 GiB
usage: 11 TiB used, 60 TiB / 71 TiB avail
pgs: 384 active+clean
According to the provided data, you should evaluate the following considerations and scenarios:
The replication size is inclusive, and once the min_size is achieved in a write operation, you receive a completion message. That means you should expect storage consumption with the minimum of min_size and maximum of the replication size.
Ceph stores metadata and logs for housekeeping purposes, obviously consuming storage.
If you do benchmark operation via "rados bench" or a similar interface with the --no-cleanup parameter, objects will be permanently stored within the cluster that consumes storage.
All the mentioned scenarios are a couple of possibilities.
I'm working on setup a Ceph cluster with Docker and image 'ceph/daemon:v3.1.0-stable-3.1-luminous-centos-7'. But after the cluster has been setup, the ceph status command never reaches HEALTH_OK. Here is my cluster's information. It has enough disk space and the network is all right.
My question are:
Why does Ceph not replicate the 'undersized' pages?
How to fix it?
Thank you very much!
➜ ~ ceph -s
cluster:
id: 483a61c4-d3c7-424d-b96b-311d2c6eb69b
health: HEALTH_WARN
Degraded data redundancy: 3 pgs undersized
services:
mon: 3 daemons, quorum pc-10-10-0-13,pc-10-10-0-89,pc-10-10-0-160
mgr: pc-10-10-0-89(active), standbys: pc-10-10-0-13, pc-10-10-0-160
mds: cephfs-1/1/1 up {0=pc-10-10-0-160=up:active}, 2 up:standby
osd: 5 osds: 5 up, 5 in
rbd-mirror: 3 daemons active
rgw: 3 daemons active
data:
pools: 6 pools, 68 pgs
objects: 212 objects, 5.27KiB
usage: 5.02GiB used, 12.7TiB / 12.7TiB avail
pgs: 65 active+clean
3 active+undersized
➜ ~ ceph osd tree
ID CLASS WEIGHT TYPE NAME STATUS REWEIGHT PRI-AFF
-1 12.73497 root default
-5 0.90959 host pc-10-10-0-13
3 hdd 0.90959 osd.3 up 1.00000 1.00000
-7 0.90959 host pc-10-10-0-160
4 hdd 0.90959 osd.4 up 1.00000 1.00000
-3 10.91579 host pc-10-10-0-89
0 hdd 3.63860 osd.0 up 1.00000 1.00000
1 hdd 3.63860 osd.1 up 1.00000 1.00000
2 hdd 3.63860 osd.2 up 1.00000 1.00000
➜ ~ ceph osd pool ls detail
pool 1 'cephfs_data' replicated size 3 min_size 2 crush_rule 0 object_hash rjenkins pg_num 8 pgp_num 8 last_change 24 flags hashpspool stripe_width 0 application cephfs
pool 2 'cephfs_metadata' replicated size 3 min_size 2 crush_rule 0 object_hash rjenkins pg_num 8 pgp_num 8 last_change 24 flags hashpspool stripe_width 0 application cephfs
pool 3 '.rgw.root' replicated size 3 min_size 2 crush_rule 0 object_hash rjenkins pg_num 13 pgp_num 13 last_change 27 flags hashpspool stripe_width 0 application rgw
pool 4 'default.rgw.control' replicated size 3 min_size 2 crush_rule 0 object_hash rjenkins pg_num 13 pgp_num 13 last_change 30 flags hashpspool stripe_width 0 application rgw
pool 5 'default.rgw.meta' replicated size 3 min_size 2 crush_rule 0 object_hash rjenkins pg_num 13 pgp_num 13 last_change 32 owner 18446744073709551615 flags hashpspool stripe_width 0 application rgw
pool 6 'default.rgw.log' replicated size 3 min_size 2 crush_rule 0 object_hash rjenkins pg_num 13 pgp_num 13 last_change 34 flags hashpspool stripe_width 0 application rgw
#itsafire This is not the solution. He is asking for solution not asking for hardware recommendation.
I'm running 8 nodes and 5 nodes multiple CEPH clusters. I always use 2 replica with multiple crush map (for SSD, SAS and 72k drives)
Why you need 3 replica if you are using a small cluster with limited resources.
Could you please explain why my solution is Recipe for disaster? You have good reputation and I'm not sure how did you get them. Maybe just replying recommendation not solution.
Create a new Pool with Size 2 and Min Size 1.
For pg-num use Ceph PG Calculator https://ceph.com/pgcalc/
It seems you created a three node cluster with different osd configurations and sizes. The standard crush rule tells ceph to have 3 copies of a PG on different hosts. If there is not enough space to spread the PGs over the three hosts, then your cluster will never be healthy.
It is always a good idea to start with a set of equally sized hosts (RAM, CPU, OSDs).
Update for discussion about cluster with size of 2 vs 3
Don't use 2 replicas. Go for 3. Ceph started out with a size default of 2. But this was changed to 3 in Ceph 0.82 (Firefly release).
Why ? Because if one drive fails you are left with only one drive containing your data. Should this drive fail too while recovery is running, then your data is gone for good.
See this thread on the ceph user mailing list
2 replicas isn't safe, no matter how big or small the cluster is. With
disks becoming larger recovery times will grow. In that window you don't
want to run on a single replica.
We have been working on restoring our Ceph cluster after losing a large number of OSDs. We have all PGs active now except for 80 PGs that are stuck in the "incomplete" state. These PGs are referencing OSD.8 which we removed 2 weeks ago due to corruption.
We would like to abandon the "incomplete" PGs as they are not restorable. We have tried the following:
Per the docs, we made sure min_size on the corresponding pools was
set to 1. This did not clear the condition.
Ceph would not let us
issue "ceph osd lost N" because OSD.8 had already been removed from
the cluster.
We also tried "ceph pg force_create_pg X" on all the
PGs. The 80 PGs moved to "creating" for a few minutes but then all
went back to "incomplete".
How do we abandon these PGs to allow recovery to continue? Is there some way to force individual PGs to be marked as "lost"?
To remove the OSD we used the procedure from the web site here:
http://docs.ceph.com/docs/jewel/rados/operations/add-or-rm-osds/#removing-osds-manual
Basically:
ceph osd crush remove 8
ceph auth del osd.8
ceph osd rm 8
Some miscellaneous data below:
djakubiec#dev:~$ ceph osd lost 8 --yes-i-really-mean-it
osd.8 is not down or doesn't exist
djakubiec#dev:~$ ceph osd tree
ID WEIGHT TYPE NAME UP/DOWN REWEIGHT PRIMARY-AFFINITY
-1 58.19960 root default
-2 7.27489 host node24
1 7.27489 osd.1 up 1.00000 1.00000
-3 7.27489 host node25
2 7.27489 osd.2 up 1.00000 1.00000
-4 7.27489 host node26
3 7.27489 osd.3 up 1.00000 1.00000
-5 7.27489 host node27
4 7.27489 osd.4 up 1.00000 1.00000
-6 7.27489 host node28
5 7.27489 osd.5 up 1.00000 1.00000
-7 7.27489 host node29
6 7.27489 osd.6 up 1.00000 1.00000
-8 7.27539 host node30
9 7.27539 osd.9 up 1.00000 1.00000
-9 7.27489 host node31
7 7.27489 osd.7 up 1.00000 1.00000
BUT, even though OSD 8 no longer exists I see still lots of references to OSD 8 in various ceph dumps and query's.
Interestingly, we do still see weird entries in the CRUSH map (should I do something about these?):
# devices
device 0 device0
device 1 osd.1
device 2 osd.2
device 3 osd.3
device 4 osd.4
device 5 osd.5
device 6 osd.6
device 7 osd.7
device 8 device8
device 9 osd.9
And for what it is worth, here is the ceph -s:
cluster 10d47013-8c2a-40c1-9b4a-214770414234
health HEALTH_ERR
212 pgs are stuck inactive for more than 300 seconds
93 pgs backfill_wait
1 pgs backfilling
101 pgs degraded
63 pgs down
80 pgs incomplete
89 pgs inconsistent
4 pgs recovery_wait
1 pgs repair
132 pgs stale
80 pgs stuck inactive
132 pgs stuck stale
103 pgs stuck unclean
97 pgs undersized
2 requests are blocked > 32 sec
recovery 4394354/46343776 objects degraded (9.482%)
recovery 4025310/46343776 objects misplaced (8.686%)
2157 scrub errors
mds cluster is degraded
monmap e1: 3 mons at {core=10.0.1.249:6789/0,db=10.0.1.251:6789/0,dev=10.0.1.250:6789/0}
election epoch 266, quorum 0,1,2 core,dev,db
fsmap e3627: 1/1/1 up {0=core=up:replay}
osdmap e4293: 8 osds: 8 up, 8 in; 144 remapped pgs
flags sortbitwise
pgmap v1866639: 744 pgs, 10 pools, 7668 GB data, 20673 kobjects
8339 GB used, 51257 GB / 59596 GB avail
4394354/46343776 objects degraded (9.482%)
4025310/46343776 objects misplaced (8.686%)
362 active+clean
112 stale+active+clean
89 active+undersized+degraded+remapped+wait_backfill
66 active+clean+inconsistent
63 down+incomplete
19 stale+active+clean+inconsistent
17 incomplete
5 active+undersized+degraded+remapped
4 active+recovery_wait+degraded
2 active+undersized+degraded+remapped+inconsistent+wait_backfill
1 stale+active+clean+scrubbing+deep+inconsistent+repair
1 active+remapped+inconsistent+wait_backfill
1 active+clean+scrubbing+deep
1 active+remapped+wait_backfill
1 active+undersized+degraded+remapped+backfilling