HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-657455681
@dbtsai that sounds great. I will leave the link at SPARK-32253
This is an automated message from the
HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-657305140
Sounds good. We should fix the flakiness. I will share it a bit later when
we're good.
This is an
HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-657301895
Actually, we should fix to make it easier to read at SPARK-32253. I will do
it a bit later to document how to read but sure I will still share some
contexts in dev mailing
HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-657145000
Thank you all guys!
This is an automated message from the Apache Git Service.
To respond to the message,
HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-656971553
Okay, `org.apache.spark.sql.SQLQueryTestSuite` already runs in parallel in a
dedicated forked JVM so it didn't decrease the time much here but I can see
these are correctly
HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-656967843
Even after splitting, SQL is still the slowest one but I think it is good
enough for now. We can mark more tests as `ExtendedSQLTest` to evenly run.
Just for the
HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-656966067
The current GitHun Actions build virtually test everything and mimics what
the current PR builder on Jenkins does. Yes, I do believe we can merge PRs if
one of GitHub
HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-656954922
retest this please
This is an automated message from the Apache Git Service.
To respond to the message,
HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-656951473
Yes, we should wait for Jenkins result because Jenkins and Github Actions
share the same script. Also, there was an issue in Jenkins and GitHub actions
about conda.
HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-656950448
@dongjoon-hyun, that's a good idea. I was thinking about splitting it too.
Let me split.
This is an
HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-656632267
retest this please
This is an automated message from the Apache Git Service.
To respond to the message,
HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-656602767
retest this please
This is an automated message from the Apache Git Service.
To respond to the message,
HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-656523303
retest this please
This is an automated message from the Apache Git Service.
To respond to the message,
HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-656467208
I reduced 3 jobs ^ but the total build time should remain same.
This is an automated message from the
HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-656456438
Ah .. conda is the problem again. Let me take a closer look.
This is an automated message from the Apache
HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-656437440
I believe now it's good to go when all the tests pass.
This is an automated message from the Apache Git
HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-656228217
Just for the record, all tests passed at here
https://github.com/HyukjinKwon/spark/runs/854407596, and it took 1h 47m 25s.
HyukjinKwon commented on pull request #29057:
URL: https://github.com/apache/spark/pull/29057#issuecomment-656149858
I already individually tested, and I am pretty sure it will show a green
light. See also https://github.com/HyukjinKwon/spark/runs/854113657
18 matches
Mail list logo