codenohup opened a new pull request, #3318:
URL: https://github.com/apache/celeborn/pull/3318
<!--
Thanks for sending a pull request! Here are some tips for you:
- Make sure the PR title start w/ a JIRA ticket, e.g. '[CELEBORN-XXXX]
Your PR title ...'.
- Be sure to keep the PR description updated to reflect all changes.
- Please write your PR title to summarize what this PR proposes.
- If possible, provide a concise example to reproduce the issue for a
faster review.
-->
### What changes were proposed in this pull request?
Some minor performance optimizations in the internal implementation
### Why are the changes needed?
During our use of Flink with Celeborn, we identified several minor
optimizations that can be made:
1. In the client side, the Flink-Celeborn client parses the
`pushDataTimeout` configuration too frequently, which is unnecessary and
cpu-intensive.
2. On the worker side, Celeborn needs to filter readers that are able for
reading. However, using Java Stream's collection operations is costly in terms
of performance.
3. Also on the worker side, Celeborn currently checks whether a reader can
continue reading by comparing the current read offset with the total file size.
This check involves retrieving the total file size, which is an expensive
operation. Since this value is constant, it should be cached in memory instead
of being fetched multiple times.
4. In the Flink’s hybrid shuffle integration, the `EndOfSegment` event
should not be bundled with data buffers. If it is, there is a risk of data
corruption or misinterpretation.
### Does this PR introduce _any_ user-facing change?
No
### How was this patch tested?
Manual test.
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
