12-11-2015 12:21 PM
I am having issue with cloudera manager as show in the picture. I have follow the restart procedure to restart the all the hadoop system but the cloudera manager still show error. My servers in this environment is using elastic IP. Please help.
12-11-2015 02:11 PM
The question marks just mean CM doesn't know all the details - it doesn't mean anything is wrong with that particular service. In my experience, this is usually just because of a hiccup in the Host Monitor (or one of the other "Cloudera Manager Services"). Click on "Cloudera Mana..." near the bottom, see if any of the roles are showing an error. Often, just restarting that role is sufficient.
01-04-2016 06:12 PM
This happens to my cluster after every start (I always start "cold", from snapshots, since I'm running on spot instances) when host monitor gives up attempting to connect to cloudera manager before cloudera manager is fully initialized (which can be confirmed/validated by inspecting host monitor logs). One solution is to make sure host with Cloudera Manager is up first (before the rest of the nodes are up). Since that is not always possible (like in my case when there is no way to control in which order your spot fleet request is fulfilled by AWS), you can just wait for cloudera manager fully initialize and then bounce cloudera-scm-agent (hard restart) on the node where host monitor is running (hard restart will force [re]-start of supervisord with all the underlying agents).