Yogesh Nath created SPARK-15578:
-----------------------------------
Summary: Run Spark jobs in Mesos sequentially
Key: SPARK-15578
URL: https://issues.apache.org/jira/browse/SPARK-15578
Project: Spark
Issue Type: Improvement
Components: Mesos, Spark Submit
Affects Versions: 1.6.1
Reporter: Yogesh Nath
Priority: Blocker
I schedule multiple jobs using Spark Dispatcher, initially, only 1 driver is
launched taking all the resources available but then after that is complete,
all other drivers are run at the same time and resources are spliced across
them.
There should be a way to specify that Jobs should be run sequentially so that
the first job in the queue is given a chance to take as many resources as it
initially requested(based on --num-executor) before other jobs in the queue are
executed.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
