Github user suryag10 commented on the issue:
https://github.com/apache/spark/pull/22433
> It is an implementation detail of k8s integration that application name
is expected to be DNS compliant ... spark does not have that requirement; and
yarn/mesos/standalone/local work without this restriction.
> The right fix in k8s integration would be to sanitize the name specified
by user/application to be compliant with its requirements. This will help not
just with thrift server, but any spark application.
As this script is common start point for all the resource
managers(k8s/yarn/mesos/standalone/local), i guess changing this to fit for all
the cases has a value add, instead of doing at each resource manager level.
Thoughts?
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
