Chapter 21. Storage Cluster Upgrade

To upgrade ceph-deploy, execute:

sudo yum install ceph-deploy

To upgrade monitors, execute the following on your monitor nodes:

ceph-deploy install <ceph-node>[<ceph-node> ...]

ceph-deploy will install the latest version of Ceph.

Restart your monitors one at a time. Give each daemon time to come up and in, rejoining the quorum before you restart the next instance. To restart a monitor, execute ceph with the restart command. Use the following syntax:

sudo /etc/init.d/ceph [options] restart mon.[id]

To upgrade OSDs, execute the following on your OSD nodes:

ceph-deploy install <ceph-node>[<ceph-node> ...]

ceph-deploy will install the latest version of Ceph.

We recommend upgrading OSDs by CRUSH hierarchy—​i.e., by failure domain or performance domain. Give each daemon time to come up and in with the cluster reaching a HEALTH_OK state before proceeding to the next CRUSH hierarchy. To restart an OSD, execute ceph with the restart command. Use the following syntax:

sudo /etc/init.d/ceph [options] restart osd.[id]

To upgrade a Ceph Object Gateway daemon, execute the following:

sudo yum install ceph-radosgw

To upgrade the Ceph Object Gateway synchronization agent, execute the following:

sudo yum install radosgw-agent

Restart each Ceph Object gateway daemon. To do so, execute the following on each host:

On RHEL 7:

sudo systemctl restart ceph-radosgw

On RHEL 6:

sudo service ceph-radosgw restart

If you are running a federated architecture, restart your sync agent(s). For data replication agents, go to the terminal and execute ctrl + c; then, execute:

radosgw-agent -c [config-file]

For metadata replication agents, go to the terminal and execute ctrl + c; then, execute:

radosgw-agent -c [config-file] --metadata-only