Github user zjffdu commented on the issue:
https://github.com/apache/zeppelin/pull/3102
Thanks @egorklimov , this is an interesting feature.
This assume all the interpreter process will generate pid file, but this
assumption is not true. There's 2 exceptions at least for now. One is
yarn-cluster mode of spark where the interpreter runs in remote node of yarn
cluster. Another is running interpreter in container which is on our roadmap.
Do you have any ideas of how to handle these 2 scenarios ?
---
