search

apache / incubator-streampark

Make stream processing easier! Easy-to-use streaming application development framework and operation platform.
https://streampark.apache.org/
Apache License 2.0
3.91k stars 1.01k forks source link

[Bug] When Flink is running in yarn mode, the number of taskmanagers and memory size cannot be accurately obtained #4114

Open wangkeshuai1 opened 1 month ago

wangkeshuai1 commented 1 month ago

Search before asking

Java Version

1.8.0_345

Scala Version

2.12.x

StreamPark Version

2.1.5

Flink Version

Flink-1.16.2

deploy mode

yarn-application

What happened

This issue is similar to #1633 . I learned from the debug source code that when I got the data of Flink Overview, the data was not accurate because Flink had just started, and the data obtained by debug was shown in the following figure.

Error Exception

No response

Screenshots

image image

Are you willing to submit PR?

Code of Conduct

wangkeshuai1 commented 1 month ago

I want to submit this PR, the idea is to delete the first judgment, because I think these information need to be obtained in real time, for example, the end time of different TM may be different, if this judgment is deleted, STARTING_CACHE will be deleted at the same time, because this cache is not referenced elsewhere, may I ask if this cache still needs to be retained?