Github user harishreedharan commented on the pull request:
https://github.com/apache/spark/pull/1310#issuecomment-50915250
@joyyoj Thanks for the explanation. This makes quite a lot of sense. I
recently added a new Dstream + an associated Flume sink to fix the issue of
receivers being hard-coded on the Flume config. Basically solves the same
issue, by telling the Spark receiver where the Flume agents are running. So
even if the executors die, they can come back and simply poll the same Flume
agents for data. In my experience, the hosts on which the agents are running
rarely change - so this solution works nicely. PR# 807 - let me know what you
think.
---
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.
---
