WiredTiger's checkpoint test, aka test/checkpoint, performs transactions on a set of parallel tables. The expectation is that any snapshot or timestamp read will see the same content in all tables.
We should further check that the tables stay in sync after recovery. To test this, simulate volume snapshot and restore as the test is running, and validate that the restored tables remain in sync with each other.
- duplicates
-
WT-7861 Write a c/python test to check checkpoint data consistency when oldest timestamp moves ahead of checkpoint timestamp
- Closed
- is related to
-
WT-7984 Fix a bug that could cause a checkpoint to omit a page of data
- Closed
-
WT-7630 Fix a history store entry remove with checkpoint reserved transaction-id
- Closed
-
WT-7964 Fix rollback to stable incorrectly not rolling back updates at snap_max
- Closed
-
WT-7965 Update connection base write generation number at the end of recovery checkpoint
- Closed
-
WT-7995 Fix the global visibility so that it cannot go beyond checkpoint visibility
- Closed
-
WT-8005 Fix a prepare commit bug that could leave the history store entry unresolved
- Closed
-
WT-7999 Fix the assert to handle an update in the middle with max stop timestamp
- Closed
- related to
-
WT-7984 Fix a bug that could cause a checkpoint to omit a page of data
- Closed