Hi,
I apologize if this question was asked before. I try to find the
answer, but in vain.
I'm running PySpark on Google Cloud Platform with Spark 2.2.0 and
YARN resource manager.
Let S1 be the set of application-ids collected via 'curl
'http://127.0.0.1:18080/api/v1/applications?status=running'; and S2 be
the application ids collected via 'yarn application -list'.
Sometimes I found S1 != S2, how could this this take place?
For those in the difference of S2 - S1 (i.e. alive YARN app, dead
Spark app), I can kill them using command 'yarn application -kill id'.
How can I kill those application in S1 - S2 (i.e. alive Spark app,
dead YARN app)? Looking not closing the SparkContext could cause this
problem. However, I'm not always able to close the context, for example
my program crash prematurely.
Tons thanks in advance!
Shuxin Yang
---------------------------------------------------------------------
To unsubscribe e-mail: user-unsubscr...@spark.apache.org