Created on 01-04-2017 12:50 PM - edited 08-19-2019 04:14 AM
Hey All,
I faced this issue in HDP 2.4 cluster running on centos.
When I run 'hive' command, it always used to halt like below:
Tried to add the proxyuser properties for hosts and groups and finally found that's not actually causing this.
I stop Spark thrift server, immediately it gives me the hive prompt and I can work with hive cli.
Anybody faced similar problems ???
Regards,
Vinay MP
Created 01-04-2017 02:00 PM
hive cli will start a Application Master on yarn. If you dont have enough resources on yarn, then AM launched by hive cli will stay in Accepted state until it gets the resources. As you mentioned stopping spark thrift server resolves this, it looks that you have short of resources on yarn. Spark thrift server will also lauch a AM on yarn and this AM will stay in RUNNING status until spark thrift server is stopped.
If this is issue with yarn resources, you should see a reason mentioned on AM launched by hive CLI.
Created 01-04-2017 02:00 PM
hive cli will start a Application Master on yarn. If you dont have enough resources on yarn, then AM launched by hive cli will stay in Accepted state until it gets the resources. As you mentioned stopping spark thrift server resolves this, it looks that you have short of resources on yarn. Spark thrift server will also lauch a AM on yarn and this AM will stay in RUNNING status until spark thrift server is stopped.
If this is issue with yarn resources, you should see a reason mentioned on AM launched by hive CLI.
Created 01-25-2017 06:19 AM
Hey @rguruvannagari,
Sorry to reply so late and taking so long time on this. Since Thrift server wasn't required for our project we decided to stop in the cluster.
And thank you for the suggestion. Now got some free time and verified.
Yarn was keeping it in the ACCEPTED state as long as memory wasn't available. Once memory is available, I can see the hive prompt as the application goes to RUNNING state.