RFE: propagate custom labels to spark nodes

[jkremser]

Currently it's possible to spawn a new app with custom metadata on it but those are not copied to the spark instances oshinko is starting.. so for instance if I run:

SECONDS=$1
oc new-app --template=oshinko-scala-spark-build-dc \
    -l app=handler-$SECONDS \
    -p SBT_ARGS=assembly \
    -p OSHINKO_CLUSTER_NAME=sparky-$SECONDS \
    -p APPLICATION_NAME=equoid-data-handler-$SECONDS \
    -p GIT_URI=https://github.com/eldritchjs/equoid-data-handler \
    -p GIT_REF=ivupdate \
    -p APP_MAIN_CLASS=io.radanalytics.equoid.DataHandler \
    -e JDG_HOST=datagrid-hotrod \
    -e JDG_PORT=11222 \
    -e WINDOW_SECONDS=$SECONDS \
    -e SLIDE_SECONDS=$SECONDS \
    -e SPARK_MASTER=spark://sparky-$SECONDS:7077 \
    -p SPARK_OPTIONS='--driver-java-options=-Dvertx.cacheDirBase=/tmp'

it will create tag app=handler-$SECONDS on the build, dc, is resources, but spark master and workers don't have this tag. Later on if I want to delete everything the previous command has created, I had to do something like this:

oc delete all -l app=handler-$SECONDS
oc delete all -l oshinko-cluster=sparky-$SECONDS

Because, the tag oshinko-cluster=sparky-$SECONDS is implicitly added to each spark pod and dc.
It would be useful to have a way to propagate the user custom labels on those spark cluster resources so that only oc delete all -l app=handler-$SECONDS would be enough,