[
https://issues.apache.org/jira/browse/MAPREDUCE-7341?focusedWorklogId=736913&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-736913
]
ASF GitHub Bot logged work on MAPREDUCE-7341:
---------------------------------------------
Author: ASF GitHub Bot
Created on: 04/Mar/22 21:30
Start Date: 04/Mar/22 21:30
Worklog Time Spent: 10m
Work Description: steveloughran commented on pull request #2971:
URL: https://github.com/apache/hadoop/pull/2971#issuecomment-1059545705
just updated with changes from sseth's review
* renamed StoreOperations to ManifestStoreOperations, set scope up.
that makes for a change which touches many classes.
* lots of other review points, all minor in comparison.
+ new DirEntry type in manifest for dest dirs only, contains
dest and status. Status is always 0, "unknown", for now.
I think based on future stats of mkdir performance, we may want to
evolve dir preparation with two options.
Probing for dest dirs in task commit. no side effects and something we
can do in parallel with the listing process. Will allow all probes for
dest dirs to be omitted from job commit. There will be duplication
in the tasks, but off the critical path/parallelised with the treewalk.
Actually attempting to create dest dirs in TaskCommit. as well as being
slightly side effecting (but no new files..) we would have to deal with
* two task commits clashing. use same recovery as job commit.
* file at dest. note and report for job commit to process.
mkdir in task is clearly more complex; I will ignore for now and
leave for a future iteration based on job stats analysis of real
world jobs.
getFileStatus is low cost and low complexity.
job commit will
1. merge dir list and status
2. those with files: delete and create (do this first)
3. those not present
one issue here though: final task commit will be slower; all previous tasks
will have repeated the operation.
will it actually speed things up?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
Issue Time Tracking
-------------------
Worklog Id: (was: 736913)
Time Spent: 26h (was: 25h 50m)
> Add a task-manifest output committer for Azure and GCS
> ------------------------------------------------------
>
> Key: MAPREDUCE-7341
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-7341
> Project: Hadoop Map/Reduce
> Issue Type: New Feature
> Components: client
> Affects Versions: 3.3.1
> Reporter: Steve Loughran
> Assignee: Steve Loughran
> Priority: Major
> Labels: pull-request-available
> Time Spent: 26h
> Remaining Estimate: 0h
>
> Add a task-manifest output committer for Azure and GCS
> The S3A committers are very popular in Spark on S3, as they are both correct
> and fast.
> The classic FileOutputCommitter v1 and v2 algorithms are all that is
> available for Azure ABFS and Google GCS, and they have limitations.
> The v2 algorithm isn't safe in the presence of failed task attempt commits,
> so we
> recommend the v1 algorithm for Azure. But that is slow because it
> sequentially lists
> then renames files and directories, one-by-one. The latencies of list
> and rename make things slow.
> Google GCS lacks the atomic directory rename required for v1 correctness;
> v2 can be used (which doesn't have the job commit performance limitations),
> but it's not safe.
> Proposed
> * Add a new FileOutputFormat committer which uses an intermediate manifest to
> pass the list of files created by a TA to the job committer.
> * Job committer to parallelise reading these task manifests and submit all the
> rename operations into a pool of worker threads. (also: mkdir, directory
> deletions on cleanup)
> * Use the committer plugin mechanism added for s3a to make this the default
> committer for ABFS
> (i.e. no need to make any changes to FileOutputCommitter)
> * Add lots of IOStatistics instrumentation + logging of operations in the
> JobCommit
> for visibility of where delays are occurring.
> * Reuse the S3A committer _SUCCESS JSON structure to publish IOStats & other
> data
> for testing/support.
> This committer will be faster than the V1 algorithm because of the
> parallelisation, and
> because a manifest written by create-and-rename will be exclusive to a single
> task
> attempt, delivers the isolation which the v2 committer lacks.
> This is not an attempt to do an iceberg/hudi/delta-lake style manifest-only
> format
> for describing the contents of a table; the final output is still a directory
> tree
> which must be scanned during query planning.
> As such the format is still suboptimal for cloud storage -but at least we
> will have
> faster job execution during the commit phases.
>
> Note: this will also work on HDFS, where again, it should be faster than
> the v1 committer. However the target is very much Spark with ABFS and GCS; no
> plans to worry about MR as that simplifies the challenge of dealing with job
> restart (i.e. you don't have to)
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
---------------------------------------------------------------------
To unsubscribe, e-mail: mapreduce-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: mapreduce-issues-h...@hadoop.apache.org
