Istvan Fajth created HDFS-13174:
-----------------------------------
Summary: hdfs mover -p /path times out after 20 min
Key: HDFS-13174
URL: https://issues.apache.org/jira/browse/HDFS-13174
Project: Hadoop HDFS
Issue Type: Bug
Components: balancer & mover
Affects Versions: 3.0.0-alpha2, 2.7.4, 2.8.0
Reporter: Istvan Fajth
Fix For: 3.1.0, 3.0.1, 2.8.4, 2.7.6
In HDFS-11015 there is an iteration timeout introduced in Dispatcher.Source
class, that is checked during dispatching the moves that the Balancer and the
Mover does. This timeout is hardwired to 20 minutes.
In the Balancer we have iterations, and even if an iteration is timing out the
Balancer runs further and does an other iteration before it fails if there were
no moves happened in a few iterations.
The Mover on the other hand does not have iterations, so if moving a path runs
for more than 20 minutes, after 20 minutes Mover will stop with the following
exception reported to the console (lines might differ as this exception came
from a CDH5.12.1 installation):
java.io.IOException: Block move timed out
at
org.apache.hadoop.hdfs.server.balancer.Dispatcher$PendingMove.receiveResponse(Dispatcher.java:382)
at
org.apache.hadoop.hdfs.server.balancer.Dispatcher$PendingMove.dispatch(Dispatcher.java:328)
at
org.apache.hadoop.hdfs.server.balancer.Dispatcher$PendingMove.access$2500(Dispatcher.java:186)
at
org.apache.hadoop.hdfs.server.balancer.Dispatcher$1.run(Dispatcher.java:956)
at
java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
at
java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
at java.lang.Thread.run(Thread.java:748)
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-dev-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-dev-h...@hadoop.apache.org
