11-15-2013 12:19 PM
I have installed cloudera manager in a Rackspace cloud server. After working for a while, I have taken image of the server and deleted it.
Right now I have re-built the server from image. I have a new Public IP assigned to the server just built. I checked and made sure that cloudera manager is installed and is in running status.
# service cloudera-scm-server status
cloudera-scm-server (pid 1894) is running...
How to restart the processes (like HIVE, Hbase, etc.) that the cloudera manager was running?
11-15-2013 01:05 PM
It sounds like your CM host IP address changed. You can tell by logging in to the CM server web UI (port 7180 by default) and looking at the Hosts tab. If you haven't gotten a heartbeat in the last 15 seconds, you probably need to update the agents to point to the new server. On each host, update /etc/cloudera-scm-agent/config.ini to point to your new CM host, then restart the agent.
11-15-2013 08:03 PM
Thanks for your solution. Your solution worked. Could you please confirm if the command below is the right command to restart agent.
sudo service cloudera-scm-agent restart
After I restart the agent, how do I restart all the services on the new server (new IP)?
11-15-2013 08:07 PM
Also I can find the host name in the Hosts tab of the CM along with the old host name. But the status of the new host says ' UNKNOWN HEALTH', How to make it active?
11-15-2013 09:15 PM
Your restart command for agents is correct.
Did your agents change host names / ip addresses as well? It sounds like your hosts page has two entries for each host, one with the new name, one with the old. In this case, you need to tell your agents to use the old host id. You don't want to use the new host id because you'll have to re-configure all of your role assignments. Edit /etc/default/cloudera-scm-agent and set CMF_AGENT_ARGS="--host_id xxx" where xxx is your old host id. You can find the old host id by clicking on your old host in the Hosts tab and looking at what is listed for Host ID.
Restart your agents after changing /etc/default/cloudera-scm-agent.
On the Hosts page, you should see all of your old hosts (the ones with roles assigned to them) have recent heartbeats and good health. You can delete the copies that have no heartbeats and no roles. If you don't see this, then let us know what you see so we can fix it.
Now Cloudera Manager is correctly in contact with all of your hosts again. You can run the restart command through the Cloudera Manager UI. You can restart an individual service or the whole cluster by clicking on the appropriate dropdown menu (small button with a triangle).