Project

General

Profile

Bug #2618

error: unable to open OSD superblock

Added by John S almost 12 years ago. Updated over 10 years ago.

Status:
Can't reproduce
Priority:
Low
Assignee:
-
Category:
-
Target version:
-
% Done:

0%

Source:
Community (user)
Tags:
Backport:
Regression:
No
Severity:
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

I am new at this.
I installed ceph.
When I do a service ceph start, mon.0 and mds.(machine name) seem ok.
When it does the osd.o part, it gets an error
ERROR: unable to open OSD superblock on /data/ceph/osd0: (2) No such file or directory
failed '/usr/bin/ceph-osd -i 0 -c /etc/ceph/ceph.conf'

Any help?
Thanks
John

ceph.conf View - ceph.conf (651 Bytes) John S, 06/25/2012 03:21 PM

History

#1 Updated by Josh Durgin almost 12 years ago

Hi John,

Did you create the /data/ceph/osd0 directory? mkcephfs doesn't do it for you because of the potential for accidentally using your root partition for osd data.
If that directory does exist, you need to initialize it by running mkcephfs -a -c /etc/ceph/ceph.conf.

Josh

#2 Updated by John S almost 12 years ago

I manually created the directory.
Then I ran the mkcephfs command.
The directory has some files in it (journal, magic, whoami, ...)

When I tried to start ceph, I got that error.
That's when I posted my question.
Sorry - I tried to follow the directions.

Any other ideas? debugging tips?

#3 Updated by Dan Mick almost 12 years ago

John, can we see your ceph.conf file? If you have time, try chatting in #ceph on irc.oftc.net as well; perhaps we can get you going interactively.

#4 Updated by John S almost 12 years ago

attaching my ceph.conf.
Can't get to IRC from work - I'll try in the evenings.
thanks

#5 Updated by Josh Durgin almost 12 years ago

  • Source changed from Development to Community (user)

I noticed an issue in your ceph.conf - you have keyring = /etc/ceph/keyring.admin in the global section, and the osd and mds section reset it, but the mon section does not, so the monitors will try to use the admin keyring file. If you add 'keyring = /etc/ceph/keyring.$name' to the mon section of your ceph.conf and re-run mkcephfs, does that help?

You could also just have 'keyring = /etc/ceph/keyring.$name' in the global section and remove the other 'keyring =' lines.

#6 Updated by John S over 11 years ago

Thanks, but that didn't help.

I did notice that drives get mounted a little weird.
Don't know if that's a problem or not.
Basically I have 1 drive for my OS and 1 that I want to use for ceph.
I created a btrfs partition on the drive and mounted it as /data.
After running the ceph setup, the drive (/dev/sdb1) is mounted as /data and as /data/ceph/osd0
Is that right?
What's the recommended file placement?

#7 Updated by Samuel Just over 10 years ago

  • Status changed from New to Can't reproduce

Also available in: Atom PDF