You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@hivemall.apache.org by "Makoto Yui (Jira)" <ji...@apache.org> on 2020/04/27 07:31:00 UTC

[jira] [Created] (HIVEMALL-294) Fix XGboost to report progress report for each iteration

Makoto Yui created HIVEMALL-294:
-----------------------------------

             Summary: Fix XGboost to report progress report for each iteration
                 Key: HIVEMALL-294
                 URL: https://issues.apache.org/jira/browse/HIVEMALL-294
             Project: Hivemall
          Issue Type: Improvement
    Affects Versions: 0.6.0
            Reporter: Makoto Yui
            Assignee: Makoto Yui
             Fix For: 0.6.2


XGBoost training could be killed by Resource Manager when timeout happens.
To fix it, we need to fix XGBoost UDTF to report progress report for each iteration.

2020-04-23 07:52:13,415 INFO [AsyncDispatcher event handler] org.apache.hadoop.mapreduce.v2.app.job.impl.TaskAttemptImpl: Diagnostics report from attempt_1585011067538_2753683_r_000000_0: AttemptID:attempt_1585011067538_2753683_r_000000_0 Timed out after 3000 secs

No progress for xgboost training for 50 min.
mapreduce.task.timeout=3000



--
This message was sent by Atlassian Jira
(v8.3.4#803005)