Github user liancheng commented on the pull request:
https://github.com/apache/spark/pull/2145#issuecomment-53517950
Hey @tnachen and @scwf, the only reason I opened this PR separately is that
we're already late for the Spark 1.1 RC release, and we need a fix quickly. We
would definitely list both of you two in the contributor list of this release :)
In this PR I used `"."` as executor side Spark home when executor URI is
provided (see
[here](https://github.com/apache/spark/pull/2145/files#diff-d425d35aa23c47a62fbb538554f2f2cfR162)
and
[here](https://github.com/apache/spark/pull/2145/files#diff-e3a5e67b8de2069ce99801372e214b8eR136)).
This should be valid because we first `cd` into the right Spark home directory
(with the `basename` trick explained
[here](https://github.com/apache/spark/pull/2145/files#diff-d425d35aa23c47a62fbb538554f2f2cfR151)).
As for testing, @tnachen you really hit the point :) I only tested this PR
with a local single node Mesos cluster, which cannot simulate the situation
that Mesos slave nodes doesn't have Spark installed in the same path as the
driver side. It would be greatly appreciated if you can help testing this PR
with your real distributed Spark over Mesos cluster. Thanks in advance!
---
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]
