[
https://issues.apache.org/jira/browse/BEAM-8382?focusedWorklogId=339688&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-339688
]
ASF GitHub Bot logged work on BEAM-8382:
----------------------------------------
Author: ASF GitHub Bot
Created on: 07/Nov/19 01:56
Start Date: 07/Nov/19 01:56
Worklog Time Spent: 10m
Work Description: jfarr commented on issue #9765: [WIP][BEAM-8382] Add
rate limit policy to KinesisIO.Read
URL: https://github.com/apache/beam/pull/9765#issuecomment-550584252
> 1. Have we try to use Kinesis's RetryPolicy and BackoffStrategy? This look
like we reengineer this feature, which already supported by Kinesis's API. Here
is the example:
https://www.programcreek.com/java-api-examples/?api=com.amazonaws.retry.RetryPolicy.
If we already tried it, then what would be the reasons not to use the
Kinesis's one ?
I'm not sure if RetryPolicy and BackoffStrategy apply to
LimitExceededException / ProvisionedThroughputExceededException but I can look
into that. If so I think it makes sense to just configure this in the Kinesis
client instead of having a BackoffRateLimitPolicy. What do you think
@aromanenko-dev and @lukecwik ?
----------------------------------------------------------------
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:
us...@infra.apache.org
Issue Time Tracking
-------------------
Worklog Id: (was: 339688)
Time Spent: 9h (was: 8h 50m)
> Add polling interval to KinesisIO.Read
> --------------------------------------
>
> Key: BEAM-8382
> URL: https://issues.apache.org/jira/browse/BEAM-8382
> Project: Beam
> Issue Type: Improvement
> Components: io-java-kinesis
> Affects Versions: 2.13.0, 2.14.0, 2.15.0
> Reporter: Jonothan Farr
> Assignee: Jonothan Farr
> Priority: Major
> Time Spent: 9h
> Remaining Estimate: 0h
>
> With the current implementation we are observing Kinesis throttling due to
> ReadProvisionedThroughputExceeded on the order of hundreds of times per
> second, regardless of the actual Kinesis throughput. This is because the
> ShardReadersPool readLoop() method is polling getRecords() as fast as
> possible.
> From the KDS documentation:
> {quote}Each shard can support up to five read transactions per second.
> {quote}
> and
> {quote}For best results, sleep for at least 1 second (1,000 milliseconds)
> between calls to getRecords to avoid exceeding the limit on getRecords
> frequency.
> {quote}
> [https://docs.aws.amazon.com/streams/latest/dev/service-sizes-and-limits.html]
> [https://docs.aws.amazon.com/streams/latest/dev/developing-consumers-with-sdk.html]
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
