[
https://issues.apache.org/jira/browse/BEAM-6612?focusedWorklogId=196088&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-196088
]
ASF GitHub Bot logged work on BEAM-6612:
----------------------------------------
Author: ASF GitHub Bot
Created on: 08/Feb/19 04:34
Start Date: 08/Feb/19 04:34
Worklog Time Spent: 10m
Work Description: ajamato commented on pull request #7764: [BEAM-6612]
Modify QueueingBeamFnDataClient to become MetricsBeamFnDataClient and make all
processElement() calls occur in parallel again.
URL: https://github.com/apache/beam/pull/7764#discussion_r254953425
##########
File path:
sdks/java/harness/src/main/java/org/apache/beam/fn/harness/control/ProcessBundleHandler.java
##########
@@ -294,7 +294,7 @@ private void
createRunnerAndConsumersForPTransformRecursively(
createRunnerAndConsumersForPTransformRecursively(
beamFnStateClient,
- queueingClient,
+ metricsClient,
entry.getKey(),
entry.getValue(),
request::getInstructionId,
Review comment:
A one line change is needed to fix the test failing right now, I'll take
care of this tomorrow morning.
try (Closeable closeTracker = stateTracker.activate()) {
// becomes
try (Closeable closeTracker = MetricContainerStepMapEnvironment.activate())
{
needs to use the Metric environment. This is because Its possible the
ElementCountFnDataReceiver which comes out of the
MetricPCollectionConsumerRegistry can be called from finish() functions which
occur on this thread. So we have to use the pattern I introduced in this PR, in
this thread as well. Which I originally thought was unnecessary
See the finish function added in CombineRunners.java:
https://github.com/apache/beam/blob/5b1700db834cf949547a20a30f3ee1270710585a/sdks/java/harness/src/main/java/org/apache/beam/fn/harness/CombineRunners.java#L176
https://github.com/apache/beam/blob/5b1700db834cf949547a20a30f3ee1270710585a/sdks/java/harness/src/main/java/org/apache/beam/fn/harness/CombineRunners.java#L103
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
Issue Time Tracking
-------------------
Worklog Id: (was: 196088)
Time Spent: 0.5h (was: 20m)
> Remove QueueingBeamFnDataClient
> -------------------------------
>
> Key: BEAM-6612
> URL: https://issues.apache.org/jira/browse/BEAM-6612
> Project: Beam
> Issue Type: New Feature
> Components: java-fn-execution
> Reporter: Alex Amato
> Assignee: Alex Amato
> Priority: Major
> Labels: triaged
> Time Spent: 0.5h
> Remaining Estimate: 0h
>
> Remove QueueingBeamFnDataClient, which made process() calls all run on the
> same thread.
> [~lcwik] and I came up with this design thinking that it was required to
> process the bundle in parallel anyways, and we would have good performance.
> However after speaking to Ken, there is no requirement for a bundle or key to
> be processed in parallel. Elements are either iterables or single elements
> which defines the needs for processing a group of elements on the same thread.
> Simply performing this change will lead to the following issues:
> (1) MetricsContainerImpl and MetricsContainer are not thread safe, so when
> the process() functions enter the metric container context, they will be
> accessing an thread-unsafe collection in parallel
> (2) An ExecutionStateTracker will be needed in every thread, So we will need
> to
> create an instance and activate it in every GrpC thread which receives a new
> element.
> (Will this get sampled properly, since the trackers will be short lived).
> (3) The SimpleExecutionStates being used will need to be thread safe as well?
> I don't think so, because I don't think that the ExecutionStateSampler
> invokes them in parallel.
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
