Github user radekg commented on the issue:
https://github.com/apache/spark/pull/9608
No matter if you was to run this in bridge or host mode, the best idea
would be to request a number of ports. In case of `HOST` network only the
`port` needs to be requested. If you can use Marathon, this is what the
`portDefinitions` would look like:
```
"portDefinitions": [
{ "port": 0, "name": "spark.driver.port", ... },
{ "port": 0, "name": "spark.blockManager.port", ... },
{ "port": 0, "name": "spark.ui.port", ... }
]
```
And then, the following can be done when submitting the job:
```
spark-submit ... -Dspark.driver.port=$PORT0
-Dspark.blockManager.port=$PORT1 -Dspark.ui.port=$PORT2 ...
```
You may have to (I have no chance to test right now) request a port for
`spark.shuffle.service.port` as it uses fixed `7337`, if you run multiple spark
instances, it's best to give it a port elected by Mesos.
The respective documentation can be found here:
- https://mesosphere.github.io/marathon/docs/ports.html
- http://spark.apache.org/docs/latest/configuration.html
Hope this helps.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
