Github user benh commented on a diff in the pull request:
https://github.com/apache/spark/pull/60#discussion_r10243692
--- Diff: docs/running-on-mesos.md ---
@@ -15,13 +15,15 @@ Spark can run on clusters managed by [Apache
Mesos](http://mesos.apache.org/). F
* `export MESOS_NATIVE_LIBRARY=<path to libmesos.so>`. This path is
usually `<prefix>/lib/libmesos.so` (where the prefix is `/usr/local` by
default, see above). Also, on Mac OS X, the library is called `libmesos.dylib`
instead of `libmesos.so`.
* `export SPARK_EXECUTOR_URI=<path to
spark-{{site.SPARK_VERSION}}.tar.gz uploaded above>`.
* `export MASTER=mesos://HOST:PORT` where HOST:PORT is the host and
port (default: 5050) of your Mesos master (or `zk://...` if using Mesos with
ZooKeeper).
-8. To run a Spark application against the cluster, when you create your
`SparkContext`, pass the string `mesos://HOST:PORT` as the master URL. In
addition, you'll need to set the `spark.executor.uri` property. For example:
+8. To run a Spark application against the cluster, when you create your
`SparkContext`, pass the string `mesos://HOST:PORT` as the master URL. In
addition, you'll need to set the `spark.executor.uri` property, and other
properties for Mesos. For example:
{% highlight scala %}
val conf = new SparkConf()
.setMaster("mesos://HOST:5050")
.setAppName("My app")
.set("spark.executor.uri", "<path to spark-{{site.SPARK_VERSION}}.tar.gz
uploaded above>")
+ .set("spark.mesos.role", "spark")
+ .set("spark.mesos.checkpoint", "true"))
--- End diff --
Recent versions of Mesos (and all versions going forward) have slaves with
checkpointing turned on by default, so Spark should get allocated resources
either way. That being said, the only real reason to use this feature in Mesos
is if you often have Spark jobs that run a long time and you'd like to be able
to restart a Mesos slave without killing the Spark job that's running. Assuming
Spark jobs finish often enough (or can deal with the occasional
failed/restarted slave) it might not be worth paying the extra overhead on the
slave to checkpoint information for Spark.
Of course, I think wiring this up and letting the end user decide is great!
---
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.
---
