[
https://issues.apache.org/jira/browse/HADOOP-17461?focusedWorklogId=792887&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-792887
]
ASF GitHub Bot logged work on HADOOP-17461:
-------------------------------------------
Author: ASF GitHub Bot
Created on: 19/Jul/22 19:42
Start Date: 19/Jul/22 19:42
Worklog Time Spent: 10m
Work Description: steveloughran commented on PR #4352:
URL: https://github.com/apache/hadoop/pull/4352#issuecomment-1189482342
I now have a branch of spark set up to use this, albeit not production ready
https://github.com/steveloughran/hadoop/tree/s3/HADOOP-17461-iostatisticsContext
* TaskMetrics includes an IOStatisticsSnapshot in its serialized data
* IOStatisticsContext is retrieved and reset at start of read/write work;
updated at end.
Based on where the read bytes/write bytes counters were read.
Having done that, I've realised that the rdds and writers are not quite the
correct place as the IOStats collect all IO on the thread, and are both the
readers and writers will be working on that thread.
The place to do it is actually in the ResultTask, with IOStatisticsSnapshot
being one of the accumulators which is sent back. the spark driver would keep
the core IOStatisticsSnapshot up to date with results from success and failure
tasks.
Issue Time Tracking
-------------------
Worklog Id: (was: 792887)
Time Spent: 6h 10m (was: 6h)
> Add thread-level IOStatistics Context
> -------------------------------------
>
> Key: HADOOP-17461
> URL: https://issues.apache.org/jira/browse/HADOOP-17461
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs, fs/azure, fs/s3
> Affects Versions: 3.3.1
> Reporter: Steve Loughran
> Assignee: Mehakmeet Singh
> Priority: Major
> Labels: pull-request-available
> Time Spent: 6h 10m
> Remaining Estimate: 0h
>
> For effective reporting of the iostatistics of individual worker threads, we
> need a thread-level context which IO components update.
> * this contact needs to be passed in two background thread forming work on
> behalf of a task.
> * IO Components (streams, iterators, filesystems) need to update this context
> statistics as they perform work
> * Without double counting anything.
> I imagine a ThreadLocal IOStatisticContext which will be updated in the
> FileSystem API Calls. This context MUST be passed into the background threads
> used by a task, so that IO is correctly aggregated.
> I don't want streams, listIterators &c to do the updating as there is more
> risk of double counting. However, we need to see their statistics if we want
> to know things like "bytes discarded in backwards seeks". And I don't want to
> be updating a shared context object on every read() call.
> If all we want is store IO (HEAD, GET, DELETE, list performance etc) then the
> FS is sufficient.
> If we do want the stream-specific detail, then I propose
> * caching the context in the constructor
> * updating it only in close() or unbuffer() (as we do from S3AInputStream to
> S3AInstrumenation)
> * excluding those we know the FS already collects.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
