Details
-
Bug
-
Resolution: Done
-
Major - P3
-
None
-
None
-
Fully Compatible
-
ALL
-
RPL A (10/09/15), Repl B (10/30/15)
-
0
Description
We need to truncate the oplog to the start of the active/failed oplog batch when we recover from a failed batch.
While database operations are idempotent, the inserting of the oplog entries are not, and need to be removed (via truncate) before we apply again, much like rollback does.
Attachments
Issue Links
- is depended on by
-
SERVER-18983 Process oplog inserts, and applying, on the secondary in parallel
-
- Closed
-
-
SERVER-20921 Add jstest for oplog truncation on recovery
-
- Closed
-