[
https://issues.apache.org/jira/browse/MAPREDUCE-318?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12745837#action_12745837
]
Arun C Murthy commented on MAPREDUCE-318:
-----------------------------------------
It's fascinating to find creative ways to kick myself after a quarter-year...
here goes:
# non-public apis being made public... javadoc warnings?
# Remove logSlowShuffle - that was only for debugging
# Cut down logging at INFO level
# Re-factor construction/parsing of shuffle request and document heavily.
# Create an object to represent the shuffle header and make it a Writable
# Fetcher
#* Rename STALLED_COPY_TIMEOUT to DEFAULT_STALLED_COPY_TIMEOUT
#* Pull sanity checks in copyMapOutput to a separate function and document
# MergeManager
#* Document rationale of why all fetch threads will never stall i.e. will allow
one thread to go past always
#* Document rationale of having unconditionalReserve and reserve (mem-to-mem
merger needs this)
# Testing
#* Test with exceptions (IOException, OOM etc.) from each of the threads:
Fetcher, EventFetcher, ShuffleScheduler etc.
----
*smile* I'm pretty sure I can't be blamed for:
{noformat}
- throws ClassNotFoundException, IOException, InterruptedException {
+ throws Throwable {
{noformat}
Please do not use this bad idiom.
> Refactor reduce shuffle code
> ----------------------------
>
> Key: MAPREDUCE-318
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-318
> Project: Hadoop Map/Reduce
> Issue Type: Improvement
> Reporter: Owen O'Malley
> Assignee: Owen O'Malley
> Attachments: HADOOP-5233_api.patch, HADOOP-5233_part0.patch,
> mapred-318-14Aug.patch, mapred-318-20Aug.patch, mapred-318-common.patch
>
>
> The reduce shuffle code has become very complex and entangled. I think we
> should move it out of ReduceTask and into a separate package
> (org.apache.hadoop.mapred.task.reduce). Details to follow.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
