[
https://issues.apache.org/jira/browse/FLINK-23593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17397280#comment-17397280
]
Piotr Nowojski commented on FLINK-23593:
Thanks for the detailed investigation [~zhuzh].
[
https://issues.apache.org/jira/browse/FLINK-23593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17397119#comment-17397119
]
Timo Walther commented on FLINK-23593:
--
Thanks for investigating this [~zhuzh]. Your explanation
[
https://issues.apache.org/jira/browse/FLINK-23593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17396636#comment-17396636
]
Zhu Zhu commented on FLINK-23593:
-
I think I find the cause of the regression.
*Cause*
The regression
[
https://issues.apache.org/jira/browse/FLINK-23593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17395983#comment-17395983
]
Zhu Zhu commented on FLINK-23593:
-
I tried the benchmarks locally before/after applying FLINK-23372 and
[
https://issues.apache.org/jira/browse/FLINK-23593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17393950#comment-17393950
]
Zhu Zhu commented on FLINK-23593:
-
>> Could the larger difference between local benchmark vs. cloud be
[
https://issues.apache.org/jira/browse/FLINK-23593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17393878#comment-17393878
]
Timo Walther commented on FLINK-23593:
--
[~zhuzh] thanks for looking into it as well. Could the
[
https://issues.apache.org/jira/browse/FLINK-23593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17393836#comment-17393836
]
Zhu Zhu commented on FLINK-23593:
-
Thanks for the updates! [~sewen]
I think your guess about *Trying to
[
https://issues.apache.org/jira/browse/FLINK-23593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17393287#comment-17393287
]
Stephan Ewen commented on FLINK-23593:
--
Here is a summary from discussing this offline with
[
https://issues.apache.org/jira/browse/FLINK-23593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17392423#comment-17392423
]
Piotr Nowojski commented on FLINK-23593:
Because of
[
https://issues.apache.org/jira/browse/FLINK-23593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17392337#comment-17392337
]
Timo Walther commented on FLINK-23593:
--
I performed a couple of benchmarks locally for the
[
https://issues.apache.org/jira/browse/FLINK-23593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17392298#comment-17392298
]
Piotr Nowojski commented on FLINK-23593:
{quote}
I'd like to understand why the regression
[
https://issues.apache.org/jira/browse/FLINK-23593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17392173#comment-17392173
]
Stephan Ewen commented on FLINK-23593:
--
I would guess that with tasks no longer sharing the slot,
[
https://issues.apache.org/jira/browse/FLINK-23593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17392091#comment-17392091
]
Zhu Zhu commented on FLINK-23593:
-
I'd like to understand why the regression happens due to FLINK-23372
[
https://issues.apache.org/jira/browse/FLINK-23593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17392073#comment-17392073
]
Timo Walther commented on FLINK-23593:
--
I'm wondering if we should simply ignore this regression.
[
https://issues.apache.org/jira/browse/FLINK-23593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=17392033#comment-17392033
]
Piotr Nowojski commented on FLINK-23593:
I've run a git bisect on the above range and tracked
15 matches
Mail list logo