[
https://issues.apache.org/jira/browse/BEAM-3645?focusedWorklogId=267467&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-267467
]
ASF GitHub Bot logged work on BEAM-3645:
----------------------------------------
Author: ASF GitHub Bot
Created on: 26/Jun/19 10:37
Start Date: 26/Jun/19 10:37
Worklog Time Spent: 10m
Work Description: robertwb commented on pull request #8872: [BEAM-3645]
add ParallelBundleManager
URL: https://github.com/apache/beam/pull/8872#discussion_r297596065
##########
File path: sdks/python/apache_beam/runners/portability/fn_api_runner.py
##########
@@ -139,6 +139,18 @@ def done(self):
self._state = self.DONE_STATE
+class _ListBuffer(list):
+ """Used to support parititioning of a list."""
+ def partition(self, n):
+ n = min(n, len(self))
Review comment:
We already need to be able to support empty groups (e.g. initially all
timers are empty, when timers are fired all "normal" inputs are empty, some
PCollections are actually empty). If there are hanging tests, we should
investigate these.
In terms of hanging tests, the key is that the worker waits for a data
stream for every InputPortOperator it has in the graph. So there must be a set
of elements (even if it is empty) to send to each of these. Maybe this
invariant is being broken somewhere.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
Issue Time Tracking
-------------------
Worklog Id: (was: 267467)
Time Spent: 12.5h (was: 12h 20m)
> Support multi-process execution on the FnApiRunner
> --------------------------------------------------
>
> Key: BEAM-3645
> URL: https://issues.apache.org/jira/browse/BEAM-3645
> Project: Beam
> Issue Type: Improvement
> Components: sdk-py-core
> Affects Versions: 2.2.0, 2.3.0
> Reporter: Charles Chen
> Assignee: Hannah Jiang
> Priority: Major
> Time Spent: 12.5h
> Remaining Estimate: 0h
>
> https://issues.apache.org/jira/browse/BEAM-3644 gave us a 15x performance
> gain over the previous DirectRunner. We can do even better in multi-core
> environments by supporting multi-process execution in the FnApiRunner, to
> scale past Python GIL limitations.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
