Retry failed ssh commands. This is especially useful during system startup when the hosts may not have yet come on-line but can be useful at other points for people with flakey connections

not mesos

when deleting security groups fails, fixing bug when using all zones but
only 1 slave.

between so wait for AWS eventual consistency to catch up.

SPARK-539 #resolve

handling Boto connection exceptions (like AuthFailure)

across multiple availability zones in order to make the cluster more
resilient to failure

are done on instances in specific security groups, this seems like a
reasonable thing to clean up.