In local mode, tez log should be with the hive log. You could check hive log
for more details.
By default, hive log is located in {java.io.tmpdir}/${user.name}/hive.log
You can check $HIVE_HOME/conf/hive-log4j.properties if you set it as other
value. Thread dump should be helpful if it is some deadlock issue.
Best Regard,
Jeff Zhang
From: Thomas Larsson
<[email protected]<mailto:[email protected]>>
Reply-To: "[email protected]<mailto:[email protected]>"
<[email protected]<mailto:[email protected]>>
Date: Wednesday, June 3, 2015 at 4:45 PM
To: "[email protected]<mailto:[email protected]>"
<[email protected]<mailto:[email protected]>>
Subject: How to debug hanging hive on tez job
Hello,
I'm not sure if this question belongs to the hive or tez community so bear with
me.
I am running some hive queries on hive configured with tez in local mode as the
execution engine.
I am using hive version 0.14 and tez version 0.5.2 (HDP 2.2.4.2).
When I run the hive query, the jobs run for a while and then get stuck, just
printing out this same monitoring line over and over (i.e no changes happen for
any of the vertices)
Map 1: -/- Map 11: 1/1 Map 12: 1/1 Map 13: -/- Map 15: -/- Map
17: 1/1 Map 18: -/- Map 19: 1/1 Map 20: -/- Map 21: -/- Map 22:
-/- Map 23: -/- Map 24: -/- Map 25: 1/1 Map 26: -/- Map 27: -/-
Map 7: -/- Reducer 10: 0/1 Reducer 14: 0/1 Reducer 16: 1/1
Reducer 3: 0(+1)/1 Reducer 4: 0/1 Reducer 5: 0/1 Reducer 6: 0/1
Reducer 8: 1/1
Do you have any suggestion as to how to find out why it gets stuck?
The query is fairly complex but the input data tables are either very small or
even empty.
There is so much in the logs that I don't know what to look for.
Taking a thread dump when this happens is a bit like looking in a haystack for
me.
Anyone?
/Thomas
