The architecture of the WiredTiger storage engine implementation means that MongoDB maintains two copies of data to guarantee durability, one in the oplog the other in the WiredTiger log (journal). It is only necessary to ensure that the content of the oplog is made durable to the current guaranteed durability point, since the oplog can be used to "recover" the content in other collections after a crash.
This work encompasses switching which content is journaled when using the WiredTiger storage engine.
There are cases where the storage engine will need to keep journaling as it does today - in any configuration where the oplog isn't enabled, so it makes sense to add this feature alongside the current method of journaling. Perhaps we need a new (or expanded) parameter to the KVEngine constructor to decide which journaling mode to use.
- depends on
-
SERVER-29212 Ensure WiredTiger checkpoints are created at valid timestamps
- Closed
- is depended on by
-
SERVER-29213 Have KVWiredTigerEngine implement StorageEngine::recoverToStableTimestamp
- Closed
- is duplicated by
-
WT-2649 Some way to indicate valid points in the WT log
- Closed
- related to
-
SERVER-35572 Always allow log archiving in WiredTiger
- Closed
-
SERVER-54567 uninitialized WiredTigerRecordStore::Params field in WiredTigerHarnessHelper::newCappedRecordStore()
- Closed