-
Hi, my production environment does not allow me to add a listener to the Spark shells. Therefore, I get the execution plan from the Spark history server API. Is it possible for Spline agent to read the downloaded execution plan instead of a listener? Thanks. |
Beta Was this translation helpful? Give feedback.
Answered by
wajda
Apr 21, 2023
Replies: 1 comment
-
Answered in AbsaOSS/spline-spark-agent#659. |
Beta Was this translation helpful? Give feedback.
0 replies
Answer selected by
liujiawen
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Answered in AbsaOSS/spline-spark-agent#659.
It is potentially possible, but that would be another Spline agent project that would interact with or integrate into a Spark History server, collect the lineage data and send it to the Spline server over the Spline Producer API.