Documentation #10957
closedOSD Manual Deployment documentation error
100%
Description
Following the long-form directions at http://ceph.com/docs/master/install/manual-deployment/ on CentOS 7 results in an OSD that cannot be controlled by 'service ceph ...'
The documentation for creating a daemon on start is for a monitor, not an OSD. As a result, the line which currently reads
sudo touch /var/lib/ceph/mon/{cluster-name}-{hostname}/sysvinit
should become
sudo touch /var/lib/ceph/osd/{cluster-name}-{osd-num}/sysvinit
And the corresponding example
sudo touch /var/lib/ceph/mon/ceph-node1/sysvinit
should become
sudo touch /var/lib/ceph/osd/ceph-0/sysvinit
sudo touch /var/lib/ceph/osd/ceph-1/sysvinit
Updated by Kyle Hutson about 9 years ago
Also, for searching purposes, when you don't perform these steps first, the result of 'service ceph start osd.0' (for example) gives the error message:
/etc/init.d/ceph: osd.0 not found (/etc/ceph/ceph.conf defines mon.node-name osd.1 osd.2 [...], /var/lib/ceph defines mon.node-name osd.1 osd.2 [...])
Updated by Kefu Chai about 9 years ago
- Status changed from New to In Progress
- Assignee set to Kefu Chai
- % Done changed from 0 to 60
Updated by Kefu Chai about 9 years ago
Kyle, thanks for reporting this issue. fix posted at https://github.com/ceph/ceph/pull/3814, pending on review.
Updated by Kefu Chai about 9 years ago
- Status changed from In Progress to Resolved
- % Done changed from 60 to 100
Updated by Kyle Hutson about 9 years ago
This is now marked as resolved, however it is still wrong.
You changed the first part ('mon' to 'osd'), but not the second part (hostname to osd-num).
Updated by Kyle Hutson about 9 years ago
This patch should correctly fix it as it is now...
--- a/doc/install/manual-deployment.rst 2015-02-27 09:44:44.000000000 -0600
+++ b/doc/install/manual-deployment.rst 2015-02-27 09:45:32.000000000 -0600
@@ -445,11 +445,12 @@
In this case, to allow the start of the daemon at each reboot you
must create an empty file like this::
- sudo touch /var/lib/ceph/osd/{cluster-name}-{hostname}/sysvinit
+ sudo touch /var/lib/ceph/osd/{cluster-name}-{osd-num}/sysvinit
For example::
- sudo touch /var/lib/ceph/osd/ceph-node1/sysvinit
+ sudo touch /var/lib/ceph/osd/ceph-0/sysvinit
+ sudo touch /var/lib/ceph/osd/ceph-1/sysvinit
Once you start your OSD, it is ``up`` and ``in``.
Updated by Kefu Chai about 9 years ago
- Status changed from Resolved to In Progress
Kyle, sorry for my mistake. i am re-openning this ticket. do you want to post a pull request to github? or may i put your name in the Author field of the commit if you allow me to post the PR on behalf of you?
Updated by Kyle Hutson about 9 years ago
I haven't really done much with git. If you want to give me credit for it, fine. If you just want to fix it and call it good, no skin off my nose.
Updated by Kefu Chai about 9 years ago
kyle, i put your name in the commit log. and a PR is posted to https://github.com/ceph/ceph/pull/3832 for review.
Updated by Kefu Chai about 9 years ago
- Status changed from In Progress to Resolved