Bug #1015
closedceph 0.26, mkcephfs : ERROR: error creating empty object store
0%
Description
Hi ,
when do mkcephfs ,if it doesn't create osdN directory manually, then ,mkcephfs fail.
1. ceph 0.26 + btrfs +ubuntu 10.04 x86_64 + ceph.conf (attached);
2. mkcephfs -c /etc/ceph/ceph.conf -a -v -k adminkeyring
3.logs:
root@ubuntu-mon0:/etc/ceph# mkcephfs -c /etc/ceph/ceph.conf -a -v -k adminkeyring
temp dir is /tmp/mkcephfs.bfucWGAKhm
preparing monmap in /tmp/mkcephfs.bfucWGAKhm/monmap
/usr/bin/monmaptool --create --clobber --add 0 172.16.35.10:6789 --add 1 172.16.35.10:6790 --add 2 172.16.35.10:6791 --print /tmp/mkcephfs.bfucWGAKhm/monmap
/usr/bin/monmaptool: monmap file /tmp/mkcephfs.bfucWGAKhm/monmap
/usr/bin/monmaptool: generated fsid be1fd1b7-e585-bcd9-7ada-c078540f67ae
epoch 1
fsid be1fd1b7-e585-bcd9-7ada-c078540f67ae
last_changed 2011-04-19 13:44:52.532136
created 2011-04-19 13:44:52.532136
0: 172.16.35.10:6789/0 mon.0
1: 172.16.35.10:6790/0 mon.1
2: 172.16.35.10:6791/0 mon.2
/usr/bin/monmaptool: writing epoch 1 to /tmp/mkcephfs.bfucWGAKhm/monmap (3 monitors)
/usr/bin/cconf -c /etc/ceph/ceph.conf -n osd.0 "user"
/usr/bin/cconf -c /etc/ceph/ceph.conf -n osd.0 "ssh path"
=== osd.0 ===
pushing conf and monmap to ubuntu-osd0
--- ssh ubuntu-osd0 "cd /etc/ceph ; ulimit c unlimited ; mkdir -p /tmp/mkcephfs.bfucWGAKhm" ssh ubuntu-osd0 "cd /etc/ceph ; ulimit -c unlimited ; /sbin/mkcephfs -d /tmp/mkcephfs.bfucWGAKhm --init-daemon osd.0"
* WARNING: Ceph is still under heavy development, and is only suitable for *
* testing and review. Do not trust it with important data. *
--
2011-04-19 22:43:04.537607 7fa4cbe50720 ** ERROR: error creating empty object store in /mnt/data/osd0/osd0: error 2: No such file or directory
failed: 'ssh ubuntu-osd0 /sbin/mkcephfs -d /tmp/mkcephfs.bfucWGAKhm --init-daemon osd.0'
root@ubuntu-mon0:/etc/ceph#
Jeff Wu.
Files
Updated by Greg Farnum about 13 years ago
- Status changed from New to Rejected
You need to create the directory yourself and this is deliberate. It prevents administrators from not fully setting up their cluster and accidentally having the OSD use the root partition when they meant to use a separate drive mounted at /osd.