Forums

Home / Forums

You need to log in to create posts and topics. Login · Register

OSD won't UP after Journal temporary missing

For sharing knowledge and for others with the same case:

Timeline:

Running cluster -> prepared node01 for shutdown -> maintenance mode -> server off -> removed one of 2 NVME cards from journal for 6 drives. -> unfortunately misplaced -> server on.

Maintenance has been turned off.

Then server was correctly turned off again, card corrected. When turned on, the disks (NVME + HDD) are visible.
OSDs which had journal on unlucky NVME do not get up. 6 disks DOWN. Reboot not helepd.

5 of 6 disks I removed from the panel, added correctly again, they are UP.

One I left for testing and learning.

# ceph osd df tree (part)

-5          58.92200         -   44 TiB   27 TiB   26 TiB   74 MiB   48 GiB   17 TiB  60.93  0.90    -              host ceph01
7    hdd    7.33600   0.90002  7.3 TiB  6.0 TiB  5.9 TiB  5.5 MiB   10 GiB  1.3 TiB  81.71  1.21  253      up          osd.7
9    hdd    5.51659   1.00000  5.5 TiB  4.0 TiB  3.9 TiB  727 KiB  7.0 GiB  1.5 TiB  72.43  1.07  172      up          osd.9
10    hdd    5.51659         0      0 B      0 B      0 B      0 B      0 B      0 B      0     0    0    down          osd.10
11    hdd    3.69730   1.00000  3.7 TiB  676 GiB  615 GiB  1.5 MiB  1.2 GiB  3.0 TiB  17.86  0.26   33      up          osd.11
12    hdd    3.69730   1.00000  3.7 TiB  657 GiB  596 GiB  1.0 MiB  1.2 GiB  3.1 TiB  17.36  0.26   26      up          osd.12
13    hdd    3.69730   1.00000  3.7 TiB  2.7 TiB  2.7 TiB   43 MiB  5.3 GiB  985 GiB  73.98  1.10  115      up          osd.13

#  ceph-volume lvm list (part)

====== osd.10 ======

[block]       /dev/ceph-8c3e6404-da2d-4ec4-a0ec-a8f8a97ec04f/osd-block-26cb0014-07d1-4632-b8cb-52c7a150f23e

block device              /dev/ceph-8c3e6404-da2d-4ec4-a0ec-a8f8a97ec04f/osd-block-26cb0014-07d1-4632-b8cb-52c7a150f23e
block uuid                Piilr5-FTGQ-Yobo-ukeP-LQ9A-Lv7R-iMviwc
cephx lockbox secret
cluster fsid              db9f5673-006e-45cb-b108-3646c60a3f31
cluster name              ceph
crush device class        None
db device                 /dev/nvme0n1p1
db uuid                   39177a90-bc28-4383-8151-c22acb336295
encrypted                 0
osd fsid                  26cb0014-07d1-4632-b8cb-52c7a150f23e
osd id                    10
osdspec affinity
type                      block
vdo                       0
devices                   /dev/sdb1

[db]          /dev/nvme0n1p1

PARTUUID                  39177a90-bc28-4383-8151-c22acb336295

# ceph-volume lvm activate --all (does not work)

--> OSD ID 28 FSID 7bb1f4d4-4d9b-416f-9e99-9467f75cf48a process is active. Skipping activation
--> OSD ID 18 FSID c1d32b73-f538-4443-b002-d7806d22a313 process is active. Skipping activation
--> OSD ID 11 FSID 380f97e1-080e-4358-8d0f-7fd343a666b6 process is active. Skipping activation
--> Activating OSD ID 10 FSID 26cb0014-07d1-4632-b8cb-52c7a150f23e
Running command: /usr/bin/chown -R ceph:ceph /var/lib/ceph/osd/ceph-10
Running command: /usr/bin/ceph-bluestore-tool --cluster=ceph prime-osd-dir --dev /dev/ceph-8c3e6404-da2d-4ec4-a0ec-a8f8a97ec04f/osd-block-26cb0014-07d1-4632-b8cb-52c7a150f23e --path /var/lib/ceph/osd/ceph-10 --no-mon-config
Running command: /usr/bin/ln -snf /dev/ceph-8c3e6404-da2d-4ec4-a0ec-a8f8a97ec04f/osd-block-26cb0014-07d1-4632-b8cb-52c7a150f23e /var/lib/ceph/osd/ceph-10/block
Running command: /usr/bin/chown -h ceph:ceph /var/lib/ceph/osd/ceph-10/block
Running command: /usr/bin/chown -R ceph:ceph /dev/dm-4
Running command: /usr/bin/chown -R ceph:ceph /var/lib/ceph/osd/ceph-10
Running command: /usr/bin/ln -snf /dev/nvme0n1p1 /var/lib/ceph/osd/ceph-10/block.db
Running command: /usr/bin/chown -R ceph:ceph /dev/nvme0n1p1
Running command: /usr/bin/chown -h ceph:ceph /var/lib/ceph/osd/ceph-10/block.db
Running command: /usr/bin/chown -R ceph:ceph /dev/nvme0n1p1
Running command: /usr/bin/systemctl enable ceph-volume@lvm-10-26cb0014-07d1-4632-b8cb-52c7a150f23e
Running command: /usr/bin/systemctl enable --runtime ceph-osd@10
Running command: /usr/bin/systemctl start ceph-osd@10
--> ceph-volume lvm activate successful for osd ID: 10
--> OSD ID 16 FSID 9360b500-1ae5-44a9-adc5-4daefb61ec1b process is active. Skipping activation
--> OSD ID 7 FSID 5f14e646-dcdd-4114-88fd-94305d2283a7 process is active. Skipping activation
--> OSD ID 9 FSID 06c5f03e-197b-4c5d-8b79-72b7b792f70e process is active. Skipping activation

# /usr/bin/systemctl status ceph-osd@10 (down)
# /usr/bin/systemctl start ceph-osd@10

022-09-27T22:07:51.852+0200 7fdd72d0f700 -1 osd.10 25534 failed to load OSD map for epoch 25212, got 0 bytes
2022-09-27T22:07:51.852+0200 7fdd72d0f700 -1 osd.10 25534 failed to load OSD map for epoch 25213, got 0 bytes
2022-09-27T22:07:51.852+0200 7fdd72d0f700 -1 osd.10 25534 failed to load OSD map for epoch 25214, got 0 bytes
2022-09-27T22:07:51.852+0200 7fdd72d0f700 -1 osd.10 25534 failed to load OSD map for epoch 25215, got 0 bytes
2022-09-27T22:07:51.852+0200 7fdd72d0f700 -1 osd.10 25534 failed to load OSD map for epoch 25216, got 0 bytes
2022-09-27T22:07:51.852+0200 7fdd72d0f700 -1 osd.10 25534 failed to load OSD map for epoch 25217, got 0 bytes
2022-09-27T22:07:51.852+0200 7fdd72d0f700 -1 osd.10 25534 failed to load OSD map for epoch 25218, got 0 bytes
2022-09-27T22:07:51.852+0200 7fdd72d0f700 -1 osd.10 25534 failed to load OSD map for epoch 25219, got 0 bytes
2022-09-27T22:07:51.852+0200 7fdd72d0f700 -1 osd.10 25534 failed to load OSD map for epoch 25220, got 0 bytes
2022-09-27T22:07:51.852+0200 7fdd72d0f700 -1 osd.10 25534 failed to load OSD map for epoch 25221, got 0 bytes
2022-09-27T22:07:51.852+0200 7fdd72d0f700 -1 osd.10 25534 failed to load OSD map for epoch 25222, got 0 bytes
2022-09-27T22:07:51.852+0200 7fdd72d0f700 -1 osd.10 25534 failed to load OSD map for epoch 25223, got 0 bytes

If I understand this correctly:
1) the Journal database was corrupted, for all drives.
2) the loss of the NVME drive results in the loss of 6 OSDs