hadoop-hdfs-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Andrew Wang <andrew.w...@cloudera.com>
Subject test-patch.sh physical OOM errors
Date Mon, 10 Feb 2014 23:40:33 GMT
Hi all,

I've seen a few precommit builds recently fail with an error like this:


I think this might be related to HADOOP-10325, which involves doing an
additional maven invocation. The error looks like the machines are running
out of physical memory.

As a bandaid fix, I lowered the MAVEN_OPTS -Xmx from 3000 to 1024 for the
HDFS precommit job. I don't set this propery locally, and neither does the
HADOOP PreCommit job. YARN's uses 1024, so I figure that's a nice
conservative value.

Anyway, let me know if we start hitting Java OOMs instead and I can up it
again, or feel free to tweak this value yourself.


  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message