Hive LLAP fails to start with RUNNING_PARTIAL state

[#1]

I'm having trouble getting Hive Interactive (LLAP) to work

I'm working on a kerberized HDP 2.5 cluster.

It seems like only 1 daemon starts, here's the error:

2016-11-03 15:50:50,084 - Marker index for start of JSON data for 'llapsrtatus' comamnd : 0

2016-11-03 15:50:50,085 - LLAP app 'llap0' in 'RUNNING_PARTIAL' state. Live Instances : '1'. Desired Instances : '4' after 221.781697989 secs.

2016-11-03 15:50:50,085 - LLAP app 'llap0' did not come up after a wait of 221.782001972 seconds.

2016-11-03 15:50:50,087 - LLAP app 'llap0' deployment unsuccessful.

What is the cause for this?

[#2 · Accepted solution]

@Nasheb Ismaily

Which YARN queue are you using ? Can you attach the HSI start logs and YARN logs for the LLAP application?

Meanwhile, :

1. if you are using 'llap' named queue, can you increase the "% of cluster capacity" slider from HIVE->configs-> HSI section, or

2. if you are using any other queue, you can decrease slider for "Maximum Total Concurrent Queries".

Save the configs and restart HSI.

[#3]

I increased the capacity and it worked, thanks!

[#4]

Also I had meet this problems, and try to increase the capacity size, it still can't work. There throws an exception:

org.apache.hadoop.yarn.server.nodemanager.containermanager.linux.privileged.PrivilegedOperationException: ExitCodeException exitCode=255.yarn-master-log.tar.gz

You can get the detail info from attachment content. It's there have other solve way. Thanks.

[#5]

I'm facing the same issue:

YARN:
Memory allocated for all YARN containers on a node = 16G
Minimum Container Size (Memory) = 2G
Maximum Container Size (VCores) = 3

Hive:
% of cluster capacity = 40%
Memory per daemon = 8192
Number of LLAP Daemons = 1 
Memory per daemon = 8192 
In-Memory Cache per Daemon = 2048 
Maximum CPUs per Daemon = 3
 

I do see this error messages on the RM UI:

Diagnostics:  Unstable Application Instance : - failed with 
component LLAP failed 'recently' 6 times (4 in startup); threshold is 5 - last failure: Failure container_e29_1492031103210_0001_01_000007 on host host1.fqdn (0): http://host1.fqdn:19888/jobhistory/logs/host1.fqdn:45454/container_e29_1492031103210_0001_01_000007/...

[#6]

if nothing worked , you can try update your openssl

[#7]

Hi guys same problem that I had. I tried many things. Finally I changed my yarn llap queue max capacity from %50 to %100 and then Hive2Interactive Server successfully started. Possible cause in my case: allocated containers exceeded llap queue max capacity.