Github user asfgit closed the pull request at:
https://github.com/apache/flink/pull/393
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabl
Github user rmetzger commented on the pull request:
https://github.com/apache/flink/pull/393#issuecomment-74867033
Merging it now
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feat
Github user rmetzger commented on the pull request:
https://github.com/apache/flink/pull/393#issuecomment-74843995
Yes. First we need to understand why exactly the performance is so poor.
Maybe its an issue we can "easily" fix.
---
If your project is set up for it, you can reply to
Github user aljoscha commented on the pull request:
https://github.com/apache/flink/pull/393#issuecomment-74843783
Yes please, go ahead. But if our Pojo stuff is really that slow we should
think about how to improve that or remove it alltogether.
---
If your project is set up for it,
Github user rmetzger commented on the pull request:
https://github.com/apache/flink/pull/393#issuecomment-74833886
I would like to merge this pull request soon.
@aljoscha, do you agree that we can investigate the performance for the
PojoComparator also when the change is merged?
Github user rmetzger commented on the pull request:
https://github.com/apache/flink/pull/393#issuecomment-74754391
No, its basically CoGrouping two "big" files (~50 and ~75 GB)
https://github.com/project-flink/flink-perf/blob/master/flink-jobs/src/main/java/com/github/projectflink
Github user aljoscha commented on the pull request:
https://github.com/apache/flink/pull/393#issuecomment-74744132
What exactly are you running? TPC-H Query 3? Maybe we should test how fast
Kryo would be with the PojoComparator.
---
If your project is set up for it, you can reply to
Github user rmetzger commented on the pull request:
https://github.com/apache/flink/pull/393#issuecomment-74712141
The pull request now also contains changes to fix two classloading issues
with the newly added `ExecutionConfig`.
Also, it fixes the Web interface for YARN.
I
Github user rmetzger commented on the pull request:
https://github.com/apache/flink/pull/393#issuecomment-74665100
Thank you for reviewing the pull request.
I've addressed your remark, added a fix for another issue (FLINK-1567) and
now I'll run the performance tests.
---
If
Github user aljoscha commented on the pull request:
https://github.com/apache/flink/pull/393#issuecomment-74505787
Looks good except for the one remark I had.
Did you run any performance tests on this? What is the overhead of using an
Avro POJO vs. a plain POJO vs. using the K
Github user aljoscha commented on a diff in the pull request:
https://github.com/apache/flink/pull/393#discussion_r24749284
--- Diff:
flink-core/src/main/java/org/apache/flink/api/common/ExecutionConfig.java ---
@@ -292,7 +360,76 @@ public void registerKryoType(Class type) {
GitHub user rmetzger opened a pull request:
https://github.com/apache/flink/pull/393
[FLINK-1417] Automatically register types with Kryo
This pull request also contains code from old pull requests for FLINK-1391,
FLINK-1392, FLINK-1395.
You can merge this pull request into a
12 matches
Mail list logo
