[
https://issues.apache.org/jira/browse/BEAM-2732?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16321322#comment-16321322
]
ASF GitHub Bot commented on BEAM-2732:
--------------------------------------
pabloem opened a new pull request #4387: [BEAM-2732] Metrics rely on
statesampler state
URL: https://github.com/apache/beam/pull/4387
This PR depends on PR #4375.
This PR is a backward-compatible change that allows metrics to rely on
statesampler state to provide the current container.
This change should enable all clients to stop tracking metrics with its own
context, and start using statesampler for that.
Benchmarks showed no impact on performance. Benchmark done using an ML
Criteo pipeline that processes 10gb.
Results Before Change (5 runs):
Mean: 1809.2s, Stdev: 227.1622768
Results With Change (5 runs):
Mean: 1758.4s, Stdev: 93.46282684
----------------------------------------------------------------
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]
> State tracking in Python is inefficient and has duplicated code
> ---------------------------------------------------------------
>
> Key: BEAM-2732
> URL: https://issues.apache.org/jira/browse/BEAM-2732
> Project: Beam
> Issue Type: Bug
> Components: sdk-py-core
> Reporter: Pablo Estrada
> Assignee: Pablo Estrada
>
> e.g logging and metrics keep state separately. State tracking should be
> unified.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
