In order to best educate a solution for SERVER-42175, we should generate some data on start up and truncation performance with respect to different sizes of OplogStones.
We should add server parameters for maximum counts of OplogStones and minimum sizes of OplogStones. Additionally, we should track:
- How long oplog sampling takes on start up
- Average amount of time spent truncating
- Cumulative length of time spent truncating
- Cumulative amount of truncate calls
3. and 4. allow us to determine the average amount of time spent truncating.
- related to
-
SERVER-42175 WiredTigerRecordStore::reclaimOplog can block for extended periods of time
- Closed
-
SERVER-55821 remove next_random_sample_size=1000 configuration in the oplog sampling code
- Closed