CephBasics - Revision history

ROMAIN ROUGNY rrougny@vub.ac.be: /* Remove OSDs */

2018-03-08T14:52:22Z

Remove OSDs

← Older revision		Revision as of 14:52, 8 March 2018
Line 34:		Line 34:
	</li>		</li>
	<li> Take the OSDs out of the cluster :		<li> Take the OSDs out of the cluster :
	Before you remove an OSD, it is usually up and in. You need to take it out of the cluster so that Ceph can begin rebalancing and copying its data to other OSDs :		Before you remove an OSD, it is usually up and in. You need to take it out of the cluster so that Ceph can begin rebalancing and copying its data to other OSDs.
	<pre>		First put its weight to 0 in the crushmap:
	ceph osd ~~out~~ {osd-num}		<pre>ceph osd crush reweight osd.{osd-num} 0.0</pre>
	</pre>
	Repeat this operation for all the OSDs on the machine.		Repeat this operation for all the OSDs on the machine.
	</li>		</li>
Line 46:		Line 45:
	</pre>		</pre>
	You should see the placement group states change from active+clean to active, some degraded objects, and finally active+clean when migration completes.		You should see the placement group states change from active+clean to active, some degraded objects, and finally active+clean when migration completes.
			Once the procedure is completed, take the osd out of the cluster:
			<pre>ceph osd out {osd-num}</pre>
	</li>		</li>
	<li> Stopping the OSDs daemons		<li> Stopping the OSDs daemons

Stephane GERARD at 11:28, 4 November 2015

2015-11-04T11:28:38Z

← Older revision		Revision as of 11:28, 4 November 2015
Line 1:		Line 1:
	{{DISPLAYTITLE:Operating a Ceph cluster}}		{{DISPLAYTITLE:Operating a Ceph cluster}}
			== Where to operate ? ==
			All operations should be done on the ceph-admin machine. In our experimental testbed, it is cephq1.wn.iihe.ac.be.
	== Check Ceph cluster status ==		== Check Ceph cluster status ==
	The command :		The command :

Stephane GERARD: /* Remove OSDs */

2015-11-03T22:42:05Z

Remove OSDs

← Older revision		Revision as of 22:42, 3 November 2015
Line 69:		Line 69:
	<pre>		<pre>
	ceph osd rm {osd-num}		ceph osd rm {osd-num}
	~~#for example~~
	~~ceph osd rm 1~~
	</pre>		</pre>
	</li>		</li>

Stephane GERARD: /* Remove OSDs */

2015-11-03T22:40:37Z

Remove OSDs

← Older revision		Revision as of 22:40, 3 November 2015
Line 64:		Line 64:
	<pre>		<pre>
	ceph auth del osd.{osd-num}		ceph auth del osd.{osd-num}
			</pre>
			</li>
			<li>Remove the OSDs
			<pre>
			ceph osd rm {osd-num}
			#for example
			ceph osd rm 1
	</pre>		</pre>
	</li>		</li>

Stephane GERARD: /* Remove OSDs */

2015-11-03T22:31:37Z

Remove OSDs

← Older revision		Revision as of 22:31, 3 November 2015
Line 54:		Line 54:
	As a result, a "ceph -s" should show the OSDs as down.		As a result, a "ceph -s" should show the OSDs as down.
	</li>		</li>
	<li> Removing the OSDs		<li> Removing the OSDs :
			<ol>
	<li>Remove OSDs from crush map :		<li>Remove OSDs from crush map :
	<pre>		<pre>
Line 65:		Line 66:
	</pre>		</pre>
	</li>		</li>
			</ol>
	</li>		</li>
	</ol>		</ol>

Stephane GERARD: /* Remove OSDs */

2015-11-03T22:29:58Z

Remove OSDs

← Older revision		Revision as of 22:29, 3 November 2015
Line 55:		Line 55:
	</li>		</li>
	<li> Removing the OSDs		<li> Removing the OSDs
	##Remove OSDs from crush map :		<li>Remove OSDs from crush map :
	<pre>		<pre>
	ceph osd crush remove osd.{osd-num}		ceph osd crush remove osd.{osd-num}
	</pre>		</pre>
	##Remove the OSD authentication key :		</li>
			<li>Remove the OSD authentication key :
	<pre>		<pre>
	ceph auth del osd.{osd-num}		ceph auth del osd.{osd-num}
	</pre>		</pre>
			</li>
	</li>		</li>
	</ol>		</ol>

Stephane GERARD: /* Remove OSDs */

2015-11-03T22:27:59Z

Remove OSDs

← Older revision		Revision as of 22:27, 3 November 2015
Line 25:		Line 25:
	== Remove OSDs ==		== Remove OSDs ==
	When you want to remove a machine that contains OSDs (for example : decommissioning of an old equipment out of warranty), there is manual procedure to follow in order to do things in a clean way and to avoid problems :		When you want to remove a machine that contains OSDs (for example : decommissioning of an old equipment out of warranty), there is manual procedure to follow in order to do things in a clean way and to avoid problems :
	#Identify the OSDs hosted by the machine with the command :		<ol>
			<li> Identify the OSDs hosted by the machine with the command :
	<pre>		<pre>
	ceph osd tree		ceph osd tree
	</pre>		</pre>
	#Take the OSDs out of the cluster :		</li>
			<li> Take the OSDs out of the cluster :
	Before you remove an OSD, it is usually up and in. You need to take it out of the cluster so that Ceph can begin rebalancing and copying its data to other OSDs :		Before you remove an OSD, it is usually up and in. You need to take it out of the cluster so that Ceph can begin rebalancing and copying its data to other OSDs :
	<pre>		<pre>
Line 35:		Line 37:
	</pre>		</pre>
	Repeat this operation for all the OSDs on the machine.		Repeat this operation for all the OSDs on the machine.
	#Monitor the data migration :		</li>
			<li> Monitor the data migration :
	Once you have taken the OSDs out of the cluster, Ceph will begin rebalancing the cluster by migrating placement groups out of the OSDs you've removed. You must follow this process with the following command :		Once you have taken the OSDs out of the cluster, Ceph will begin rebalancing the cluster by migrating placement groups out of the OSDs you've removed. You must follow this process with the following command :
	<pre>		<pre>
Line 41:		Line 44:
	</pre>		</pre>
	You should see the placement group states change from active+clean to active, some degraded objects, and finally active+clean when migration completes.		You should see the placement group states change from active+clean to active, some degraded objects, and finally active+clean when migration completes.
	#Stopping the OSDs daemons		</li>
			<li> Stopping the OSDs daemons
	After you take an OSD out of the cluster, it may still be running. That is, the OSD may be up and out. You must stop your OSD before you remove it from the configuration :		After you take an OSD out of the cluster, it may still be running. That is, the OSD may be up and out. You must stop your OSD before you remove it from the configuration :
	<pre>		<pre>
Line 49:		Line 53:
	(Repeat the last command for all the OSDs on the machine.)		(Repeat the last command for all the OSDs on the machine.)
	As a result, a "ceph -s" should show the OSDs as down.		As a result, a "ceph -s" should show the OSDs as down.
	#Removing the OSDs		</li>
			<li> Removing the OSDs
	##Remove OSDs from crush map :		##Remove OSDs from crush map :
	<pre>		<pre>
Line 58:		Line 63:
	ceph auth del osd.{osd-num}		ceph auth del osd.{osd-num}
	</pre>		</pre>
			</li>
			</ol>

Stephane GERARD: /* Remove OSDs */

2015-11-03T22:20:54Z

Remove OSDs

← Older revision		Revision as of 22:20, 3 November 2015
Line 53:		Line 53:
	<pre>		<pre>
	ceph osd crush remove osd.{osd-num}		ceph osd crush remove osd.{osd-num}
			</pre>
			##Remove the OSD authentication key :
			<pre>
			ceph auth del osd.{osd-num}
	</pre>		</pre>

Stephane GERARD: /* Remove OSDs */

2015-11-03T22:17:40Z

Remove OSDs

← Older revision		Revision as of 22:17, 3 November 2015
Line 48:		Line 48:
	</pre>		</pre>
	(Repeat the last command for all the OSDs on the machine.)		(Repeat the last command for all the OSDs on the machine.)
			As a result, a "ceph -s" should show the OSDs as down.
			#Removing the OSDs
			##Remove OSDs from crush map :
			<pre>
			ceph osd crush remove osd.{osd-num}
			</pre>

Stephane GERARD: /* Remove OSDs */

2015-11-03T22:11:28Z

Remove OSDs

← Older revision		Revision as of 22:11, 3 November 2015
Line 41:		Line 41:
	</pre>		</pre>
	You should see the placement group states change from active+clean to active, some degraded objects, and finally active+clean when migration completes.		You should see the placement group states change from active+clean to active, some degraded objects, and finally active+clean when migration completes.
			#Stopping the OSDs daemons
			After you take an OSD out of the cluster, it may still be running. That is, the OSD may be up and out. You must stop your OSD before you remove it from the configuration :
			<pre>
			ssh {osd-host}
			/etc/init.d/ceph stop osd.{osd-num}
			</pre>
			(Repeat the last command for all the OSDs on the machine.)