When upgrading HDP cluster to 3.1.0 version -Restarting yarn DNS resgistry failed due to ERROR 'resource_management.core.exceptions.ExecutionFailed: Execution of 'ambari-sudo.sh -H -E /usr/hdp/22.214.171.124-78/hadoop-yarn/bin/yarn --config /usr/hdp/126.96.36.199-78/hadoop/conf --daemon start registrydns' returned 1. ERROR: Cannot set priority of registrydns process 123223'.
I ignored it and proceeded with the upgrade. But at Finalize Pre check the task failed and the only option available is to delete the unhealthy and host and proceed with the upgrade.
How can I resolve this issue.
Can you please check what is the port for the "RegistryDNS Bind Port" ? Is it free of being used by some other process ? I am assuming that port is 53 (please change the port in the following command to verify if that port is free os being used)?
# netstat -tnlpa | grep 53
If possible then can you try changing the port to something else and then see if that works? And check for "RegistryDNS Bind Port"
Ambari UI --> Yarn --> Configs --> Advanced (tab) --> Registry
Example: RegistryDNS Bind Port = 1553
Thank you for your reply, I was able to successfully restart Yarn DNS Registry but the upgrade still shows the same error - unhealthy host.
I checked the host component state in Ambari database and found that the Yarn registry is in Failed state.
YARN_REGISTRY_DNS | UNKNOWN | STARTED | 15 | YARN | FAILED