Revision 82ae1f0aca9c00fddba130c144adfe0777172cc8 authored by Tathagata Das on 15 May 2017, 17:46:38 UTC, committed by Shixiong Zhu on 15 May 2017, 17:46:45 UTC
## What changes were proposed in this pull request? StateStore.abort() should do a best effort attempt to clean up temporary resources. It should not throw errors, especially because its called in a TaskCompletionListener, because this error could hide previous real errors in the task. ## How was this patch tested? No unit test. Author: Tathagata Das <tathagata.das1565@gmail.com> Closes #17958 from tdas/SPARK-20716. (cherry picked from commit 271175e2bd0f7887a068db92de73eff60f5ef2b2) Signed-off-by: Shixiong Zhu <shixiong@databricks.com>
1 parent 0bd918f
| File | Mode | Size |
|---|---|---|
| docker.properties.template | -rw-r--r-- | 987 bytes |
| fairscheduler.xml.template | -rw-r--r-- | 1.1 KB |
| log4j.properties.template | -rw-r--r-- | 2.0 KB |
| metrics.properties.template | -rw-r--r-- | 7.1 KB |
| slaves.template | -rw-r--r-- | 865 bytes |
| spark-defaults.conf.template | -rw-r--r-- | 1.3 KB |
| spark-env.sh.template | -rwxr-xr-x | 3.6 KB |
Computing file changes ...
