Details
-
Umbrella
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
-
None
-
None
Description
This umbrella summarizes all issues related with checkpointing and task restarting to archieve fault tolerance on the job level.
Attachments
Issue Links
- relates to
-
-
- Open
-
1.
|
Integrity of checkpointed data |
|
Resolved | Suraj Menon |
2.
|
Make configurable checkpointing |
|
Resolved | Suraj Menon |
3.
|
BSPTask should periodically ping its parent. |
|
Resolved | Suraj Menon |
4.
|
Chainable computations for fault tolerance |
|
Resolved | Thomas Jungblut |
5.
|
BSP Peer should have the ability to start with a non-zero superstep from a partition of checkpointed message for that task ID, attempt ID |
|
Resolved | Unassigned |
6.
|
For configurable number of attempts, BSPMaster should direct groomserver to run the recovery task on failure. |
|
Resolved | Suraj Menon |
7.
|
Add documentation to fault tolerant job processing |
|
Open | Unassigned |
8.
|
Implement Checkpointing service in Hama |
|
Resolved | Suraj Menon |
9.
|
Handle counters during task recovery |
|
Open | Suraj Menon |
10.
|
Recover tasks on failure of groom server |
|
Open | Suraj Menon |
11.
|
Rewrite the Examples and GraphJobRunner to the Superstep API |
|
Resolved | Unassigned |
12.
|
Confined recovery |
|
Open | Unassigned |