config spark-shell to use local ip as driver host #2
Conversation
| cd /usr/local/spark | ||
| ./bin/spark-shell \ | ||
| --master spark://${SPARK_MASTER_PORT_7077_TCP_ADDR}:${SPARK_MASTER_ENV_SPARK_MASTER_PORT} \ | ||
| -i ${SPARK_LOCAL_IP} \ |
There was a problem hiding this comment.
I can't find any pointer to this option for spark-shell ? Can you help me to understand what does this option is for ?
Same option is used for launching workers using spark-class but I can't find any clue either.
There was a problem hiding this comment.
It's from some instructions on running spark shell I guess. I don't really remember
by default shell communicate its internal hostname to worker so when worker try to communicate with shell, since its hostname is internal, the worker cannot resolve the hostname. using shell local ip as driver host solve this problem. see Networking documentation of spark configuration: http://spark.apache.org/docs/latest/configuration.html#networking + fix small typo to rename start-worker to start-worker.sh
|
So your changes makes it all work without removing alias? |
|
Yes, it seems like this solve the problem. I you can test this on your side to validate that would be cool ;) |
|
And I write a tutorial referencing your work here: https://gist.github.com/clakech/4a4568daba1ca108f03c If you want to setup a spark cluster storing data into cassandra using docker, this is my 2 cents ;) |
by default shell communicate its internal hostname to worker
so when worker try to communicate with shell, since its hostname is
internal, the worker cannot resolve the hostname.
using shell local ip as driver host solve this problem.
see Networking documentation of spark configuration:
http://spark.apache.org/docs/latest/configuration.html#networking