## What changes were proposed in this pull request? - Fixes the issue with the frameworkId being recovered by checkpointed data overwriting the one sent by the dipatcher. - Keeps submission driver id as the only index for all data structures in the dispatcher. Allocates a different task id per driver retry to satisfy the mesos requirements. Check the relevant ticket for the details on that. ## How was this patch tested? Manually tested this with DC/OS 1.10. Launched a streaming job with checkpointing to hdfs, made the driver fail several times and observed behavior: ![image](https://user-images.githubusercontent.com/7945591/30940500-f7d2a744-a3e9-11e7-8c56-f2ccbb271e80.png) ![image](https://user-images.githubusercontent.com/7945591/30940550-19bc15de-a3ea-11e7-8a11-f48abfe36720.png) ![image](https://user-images.githubusercontent.com/7945591/30940524-083ea308-a3ea-11e7-83ae-00d3fa17b928.png) ![image](https://user-images.githubusercontent.com/7945591/30940579-2f0fb242-a3ea-11e7-82f9-86179da28b8c.png) ![image](https://user-images.githubusercontent.com/7945591/30940591-3b561b0e-a3ea-11e7-9dbd-e71912bb2ef3.png) ![image](https://user-images.githubusercontent.com/7945591/30940605-49c810ca-a3ea-11e7-8af5-67930851fd38.png) ![image](https://user-images.githubusercontent.com/7945591/30940631-59f4a288-a3ea-11e7-88cb-c3741b72bb13.png) ![image](https://user-images.githubusercontent.com/7945591/30940642-62346c9e-a3ea-11e7-8935-82e494925f67.png) ![image](https://user-images.githubusercontent.com/7945591/30940653-6c46d53c-a3ea-11e7-8dd1-5840d484d28c.png) Author: Stavros Kontopoulos <st.kontopoulos@gmail.com> Closes #19374 from skonto/fix_retry.