[LIVY-550] quick failure when error happens in spark-submit process

[Jassy1994]

What changes were proposed in this pull request?

support batch session a quick failure when error happens in spark-submit process

How was this patch tested?

submit a batch session with wrong username or queuename，observe session state in livy ui

[codecov-io]

Codecov Report

Merging #140 into master will increase coverage by 0.18%. The diff coverage is 100%.

@@             Coverage Diff              @@
##             master     #140      +/-   ##
============================================
+ Coverage     68.34%   68.52%   +0.18%     
+ Complexity      895      893       -2     
============================================
  Files           100      100              
  Lines          5604     5605       +1     
  Branches        840      840              
============================================
+ Hits           3830     3841      +11     
+ Misses         1225     1213      -12     
- Partials        549      551       +2

Impacted Files	Coverage Δ	Complexity Δ
...la/org/apache/livy/server/batch/BatchSession.scala	85.55% <100%> (+0.16%)	13 <0> (ø)	:arrow_down:
...ain/java/org/apache/livy/rsc/driver/RSCDriver.java	77.96% <0%> (-1.28%)	41% <0%> (-1%)
...main/java/org/apache/livy/rsc/ContextLauncher.java	66.01% <0%> (+1.94%)	13% <0%> (ø)	:arrow_down:
...scala/org/apache/livy/repl/SparkRInterpreter.scala	60.69% <0%> (+2.31%)	14% <0%> (ø)	:arrow_down:
...c/src/main/java/org/apache/livy/rsc/RSCClient.java	75.15% <0%> (+2.48%)	20% <0%> (+1%)	:arrow_up:
...cala/org/apache/livy/scalaapi/ScalaJobHandle.scala	52.94% <0%> (+2.94%)	7% <0%> (ø)	:arrow_down:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 02550f7...135b0db. Read the comment docs.

[alex-the-man]

Without this fix, the app state should still transit to FAILED after 2 minutes with default configuration. Is this fix intended to surface the error faster?

[Jassy1994]

Without this fix, the app state should still transit to FAILED after 2 minutes with default configuration. Is this fix intended to surface the error faster?

I am sorry to have a late reply, yes, app state will transit to FAILED in livy.server.yarn.app-leakage.check-timeout, in my situation, some users submit wrong params but they dont realize, and in livy ui, their session is starting so we can not correct in time.so I think a quick failure is more friendly to users.