GitHub user squito opened a pull request:
https://github.com/apache/spark/pull/6420
[SPARK-7829][CORE][WIP] multiple attempts of shuffle tasks is OK
fix for one of the retry issues -- data files are appended to, but the
index file always points to the beginning of the file. This would lead to an
apparently corrupted data file during the shuffle fetch.
for now this is just for discussion, not meant to merge until we flesh out
some other options more. Even if in master we make larger changes to deal w/
stage retry, this is small enough to potentially be backported to earlier
branches (maybe without some of the hokey test code).
cc @JoshRosen
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/squito/spark SPARK-7829
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/spark/pull/6420.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #6420
----
commit f83b66f502de8f2f5c728f09722f4f5cde5af30e
Author: Imran Rashid <[email protected]>
Date: 2015-05-26T20:42:06Z
shuffle data file is appended, so index file needs to know correct start on
multiple attempts
----
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
