Add Ceph OSDs

IMPORTANT: This document is addressing how to add an OSD when the OSD auto-discovery fails to add in new drives.

Check to ensure you have OSD auto-discovery enabled.

ncn-s00(1/2/3)# ceph orch ls osd

Example output:

NAME                       RUNNING  REFRESHED  AGE  PLACEMENT  IMAGE NAME                        IMAGE ID
osd.all-available-devices      9/9  4m ago     3d   *          registry.local/ceph/ceph:v15.2.8  5553b0cb212c

NOTE Ceph version 15.2.x and newer will utilize the ceph orchestrator to add any available drives on the storage nodes to the OSD pool. The process below is in the event that the orchestrator did not add the available drives into the cluster

Prerequisites

This procedure requires administrative privileges and will require at least two windows.

Procedure

  1. In the first window, log in as root on the first master node (ncn-m001).

    ssh ncn-m001

  2. Watch the status of the cluster to monitor the progress of the drives being added.

    The following example shows only six OSDs in use.

    watch -n 10 ceph -s

    Example output:

      cluster:
    id: 5b359a58-e6f7-4f0c-98b8-f528f620896a
    health: HEALTH_OK

  services:
    mon: 3 daemons, quorum ncn-s001,ncn-s002,ncn-s003 (age 20h)
    mgr: ncn-s003(active, since 9d), standbys: ncn-s001, ncn-s002
    mds: cephfs:1 {0=ncn-s002=up:active} 2 up:standby
    osd: 18 osds: 18 up (since 20h), 18 in (since 9d)
    rgw: 3 daemons active (ncn-s001.rgw0, ncn-s002.rgw0, ncn-s003.rgw0)

  data:
    pools:   11 pools, 220 pgs
    objects: 826.23k objects, 944 GiB
    usage:   2.0 TiB used, 61 TiB / 63 TiB avail
    pgs:     220 active+clean

  io:
    client:   1.7 KiB/s rd, 12 MiB/s wr, 1 op/s rd, 1.32k op/s wr

  3. In the second window, log into ncn-s00(1/2/3) or an ncn-m node and fail over the mgr process.

    1. There is an issue where orchestration tasks can get hung up and the failover will clear that up.
    ceph mgr fail $(ceph mgr dump | jq -r .active_name)

  4. In the second window, list your available drives on the node(s) where the OSDs are missing

    The following example is utilizing ncn-s001. Ensure the correct host for the situation is used.

    ceph orch device ls

    Example output:

    ceph orch device ls ncn-s001
Hostname  Path      Type  Serial                Size   Health   Ident  Fault  Available
ncn-s001  /dev/sdb  hdd   f94bd091-cc25-476b-9  48.3G  Unknown  N/A    N/A    No

    NOTE The drive in question is reporting available. The following steps are going to erase that drive so PLEASE make sure to verify that drive is not being used.

    podman ps

    Example output:

    CONTAINER ID  IMAGE                             COMMAND               CREATED                 STATUS                     PORTS   NAMES
596d1c235da8  registry.local/ceph/ceph:v15.2.8  -n client.rgw.sit...  Less than a second ago  Up Less than a second ago          ceph-11d5d552-cfac-11eb-ab69-fa163ec012bf-rgw.site1.ncn-s001.oztynu
eecfac35fe7c  registry.local/ceph/ceph:v15.2.8  -n mon.ncn-s001 -...  2 seconds ago           Up 2 seconds ago                   ceph-11d5d552-cfac-11eb-ab69-fa163ec012bf-mon.ncn-s001
3140f5062945  registry.local/ceph/ceph:v15.2.8  -n mgr.ncn-s001.b...  17 seconds ago          Up 17 seconds ago                  ceph-11d5d552-cfac-11eb-ab69-fa163ec012bf-mgr.ncn-s001.bfdept
3d25564047e1  registry.local/ceph/ceph:v15.2.8  -n mds.cephfs.ncn...  3 days ago              Up 3 days ago                      ceph-11d5d552-cfac-11eb-ab69-fa163ec012bf-mds.cephfs.ncn-s001.juehkw
4ebd6db27d08  registry.local/ceph/ceph:v15.2.8  -n osd.2 -f --set...  4 days ago              Up 4 days ago                      ceph-11d5d552-cfac-11eb-ab69-fa163ec012bf-osd.2
96c6e11677f0  registry.local/ceph/ceph:v15.2.8  -n client.crash.n...  4 days ago              Up 4 days ago                      ceph-11d5d552-cfac-11eb-ab69-fa163ec012bf-crash.ncn-s001

    If you find an Running OSD container then we should assume that the drive is being used or might have critical data on it. If you know this to 100% not be the case (example a rebuild), then you can proceed.

    Repeat this step for all drives on the storage node(s) that have unused storage which should be added to Ceph.

    ceph orch device zap ncn-s001 /dev/sdb (optional --force)

    Proceed to the next step after all of the OSDs have been added to the storage nodes.

  5. In the first window, check how many OSDs are available.

    The following example shows 18 OSDs in use.

    cluster:
id: 5b359a58-e6f7-4f0c-98b8-f528f620896a
health: HEALTH_ERR
Degraded data redundancy (low space): 3 pgs backfill_toofull
services:
mon: 3 daemons, quorum ncn-s001,ncn-s002,ncn-s003 (age 20h)
mgr: ncn-s003(active, since 9d), standbys: ncn-s001, ncn-s002
mds: cephfs:1 {0=ncn-s002=up:active} 2 up:standby
{0=ncn-m002=up:active}
2 up:standby
osd: 18 osds: 18 up (since 20h), 18 in (since 9d)
rgw: 3 daemons active (ncn-s001.rgw0, ncn-s002.rgw0, ncn-s003.rgw0)
data:
pools: 11 pools, 204 pgs
objects: 70.98k objects, 241 GiB
usage: 547 GiB used, 62 TiB / 63 TiB avail
pgs: 39582/212949 objects misplaced (18.588%)
163 active+clean
36 active+remapped+backfill_wait
3 active+remapped+backfill_wait+backfill_toofull
2 active+remapped+backfilling
io:
client: 20 MiB/s wr, 0 op/s rd, 807 op/s wr
recovery: 559 MiB/s, 187 objects/s

    ceph orch ps --daemon_type osd ncn-s001

    Example output:

    NAME   HOST      STATUS        REFRESHED  AGE  VERSION  IMAGE NAME                        IMAGE ID      CONTAINER ID
osd.2  ncn-s001  running (4d)  20s ago    4d   15.2.8   registry.local/ceph/ceph:v15.2.8  5553b0cb212c  4ebd6db27d08

  6. Reset the pool quotas.

    This step is only necessary when the cluster capacity has increased.

    ncn-s00(1/2/3)# source /srv/cray/scripts/common/fix_ansible_inv.sh
ncn-s00(1/2/3)# fix_inventory
ncn-s00(1/2/3)# source /etc/ansible/boto3_ansible/bin/activate
ncn-s00(1/2/3)# ansible-playbook /etc/ansible/ceph-rgw-users/ceph-pool-quotas.yml
ncn-s00(1/2/3)# deactivate