View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0017818 | CentOS-8 | general | public | 2020-10-28 15:41 | 2020-10-28 15:41 |
| Reporter | techno10 | Assigned To | |||
| Priority | normal | Severity | major | Reproducibility | always |
| Status | new | Resolution | open | ||
| Product Version | 8.2.2004 | ||||
| Summary | 0017818: systemctl restart ceph.osd@X results in issue | ||||
| Description | Running "systemctl restart ceph-osd@3" it complains about /var/lib/ceph/osd/ceph-3 missing.. and it is BUT if I do the following: [root@node1 osd]# pwd /var/lib/ceph/osd [root@node1 osd]# ln -s ../4242410a-e8d8-11ea-a7d6-001b21d6d88c/osd.3 it resolves the issue.. It seems like the systemctl script is expecting the "old" location but the new version of ceph puts these files in the FSID of the cluster (the cluster ID) AND also expects the osd name and not the "ceph-3" name (ie: osd.3 vs ceph-3). I'm guessing some package needs to be updated to match up ceph somehow? Also when it actually does restart with my work around the process looks completely different: ceph 3278 3251 1 10:04 ? 00:01:12 /usr/bin/ceph-osd -n osd.0 -f --setuser ceph --setgroup ceph --default-log-to-file=false --default-log-to-stderr=true --default-log-stderr-prefix=debug ceph 3400 3381 0 10:04 ? 00:00:37 /usr/bin/ceph-osd -n osd.2 -f --setuser ceph --setgroup ceph --default-log-to-file=false --default-log-to-stderr=true --default-log-stderr-prefix=debug ceph 3454 3423 0 10:04 ? 00:00:32 /usr/bin/ceph-osd -n osd.1 -f --setuser ceph --setgroup ceph --default-log-to-file=false --default-log-to-stderr=true --default-log-stderr-prefix=debug ceph 14467 1 0 10:42 ? 00:00:27 /usr/bin/ceph-osd -f --cluster ceph --id 3 --setuser ceph --setgroup ceph you can see the last entry has different params.. i'm guessing it'll be deprecated soon? | ||||
| Steps To Reproduce | install ceph cluster and add 2 OSDs.. stop one and try to start it. I have the latest everything from centos 8 (including ceph).. I'm using this repo: CentOS-Ceph-Octopus.repo [root@node1 ceph-3]# cat /etc/redhat-release CentOS Linux release 8.2.2004 (Core) | ||||
| Additional Information | you may see something like this in the messages file: Oct 28 10:34:20 node1 ceph-osd[13034]: 2020-10-28T10:34:20.270-0400 7f2b61b0ff40 -1 auth: unable to find a keyring on /var/lib/ceph/osd/ceph-3/keyring: (2) No such file or directory Oct 28 10:34:20 node1 ceph-osd[13034]: 2020-10-28T10:34:20.270-0400 7f2b61b0ff40 -1 AuthRegistry(0x5644f83f6940) no keyring found at /var/lib/ceph/osd/ceph-3/keyring, disabling cephx Oct 28 10:34:20 node1 ceph-osd[13034]: 2020-10-28T10:34:20.271-0400 7f2b61b0ff40 -1 auth: unable to find a keyring on /var/lib/ceph/osd/ceph-3/keyring: (2) No such file or directory Oct 28 10:34:20 node1 ceph-osd[13034]: 2020-10-28T10:34:20.271-0400 7f2b61b0ff40 -1 AuthRegistry(0x7fffa52bc210) no keyring found at /var/lib/ceph/osd/ceph-3/keyring, disabling cephx Again, the keyring exists but it's under the FSID/osd.X directory. | ||||
| Tags | ceph | ||||
