Project

General

Profile

Actions
Copy link

Bug #23124

closed

Status of OSDs are not showing properly after disabling ceph.target and ceph-osd.target

Added by Debashis Mondal about 6 years ago. Updated about 6 years ago.

Status:
Won't Fix
Priority:
Normal
Assignee:
-
Category:
-
Target version:
-
% Done:

0%

Source:
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
Ceph - v10.2.7
ceph-qa-suite:
Component(RADOS):
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

Execution Steps:
----------------
1. # ceph osd tree [ceph is in running state]
2. # systemctl disable ceph.target
3. # systemctl disable ceph-osd.target
4. # reboot
5. # ceph osd tree
6. # systemctl | grep ceph
7. # ps -ax | grep ceph

Observation:
-------------

It is observed that ceph osd tree is showing that OSDs are in "up" state , whereas osd daemon services are not running and no process is created for ceph-osd after disabling ceph.target and ceph-osd.target

Workaround:
------------

        # systemctl start ceph-osd@<osd-id>
        # systemctl stop ceph-osd@<osd-id>
[ Note : After starting the service again, osd tree gives proper output But this is not the proper solution ] # ceph osd tree

Files

CEPH_OSD_Status_Observation.txt (5.87 KB) CEPH_OSD_Status_Observation.txt Console log for details Debashis Mondal, 02/26/2018 04:03 AM
Actions
Copy link
 #1

Updated by Jason Dillaman about 6 years ago

  • Project changed from rbd to RADOS
Actions
Copy link
 #2

Updated by Nathan Cutler about 6 years ago

  • Description updated (diff)
Actions
Copy link
 #3

Updated by Nathan Cutler about 6 years ago

The ceph.target and ceph-osd.target cannot be used this way. Assuming ceph-disk is being used, the OSDs are brought up by the udev rules, regardless of the enabled status of "ceph.target" and "ceph-osd.target". I believe you can disable individual OSD units, though, i.e. "systemctl disable ceph-osd@<id>.service" and possibly (depending on the version of systemd that is installed) even "systemctl disable ceph-osd@\*.service" to disable them all.

However, if you really want to be sure it's better to use ceph-disk directly. See "ceph-disk --help" for details, but these commands should get you started:

ceph-disk list --help
ceph-disk deactivate --help
ceph-disk activate --help
Actions
Copy link
 #4

Updated by Josh Durgin about 6 years ago

  • Status changed from New to Won't Fix

As Nathan explained, this isn't how the targets are meant to work.

Actions
Copy link
 #5

Updated by Debashis Mondal about 6 years ago

As OSDs are brought up by the udev rules, regardless of the enabled status of "ceph.target" and "ceph-osd.target" hence by disabling ceph.target or ceph-osd.target the status of the OSD should not be shown up where the service regarding the OSD is already stopped and inactive.

So I think we need to look at this issue again

Actions
Copy link

Also available in: Atom PDF