When adding a new worker node with Datanode, NodeManager and Impalad cloudera manager wants to restart stale services due to the topology.map file.
We are deploying a single DC / Rack cluster so there isn't any need for us to divide into racks, all of them are in /default.
Is there any way to avoid this behavior?
And are there any negative effects ( if only using /default rack ) to not restarting the stale services when asked to update topology.map file?
The node seems to be in both YARN and HDFS node list and we can see that jobs and blocks are being copied to the host.
Can you tell me what version of CM are you using? I don't recall seeing this in recent versions.
If you only have one rack, there shouldn't be any negative inpact by not restarting stale services.
A restart should not be necessary if only the topology.map needs updating.
To refresh HDFS and YARN, refresh the cluster with the following:
A refresh does not restart any services.
Next, deploy Client configuration with steps 1 and 2 as described here:
Let us know if services still show as stale and what version of Cloudera Manager you are using if HDFS and YARN still show a stale configuration.
We are using cloudera manager with CDH 5.8.3.
Thanks for the update!
I can see now that YARN and HDFS can be refreshed without restarting the cluster.
However some services like SPARK2, HIVE, HUE and OOZIE still have the old topology.map file and cloudera manager wants to restart stale services.
Is it safe to wait until we have server maintenance or something and do the restart?