Standalone node and --repair behavior must be considered here. There are some cases where index builds are discarded/cleared, without starting up a MultiIndexBlock, and the tables must be cleared, too. We might not have a use case implemented yet, but there will be and it would be good to add the functionality here so that it is ready to use. Perhaps extend the logic that clears in-progress builds' persisted state.
- is depended on by
-
SERVER-37952 Restart index builds from persisted recovery points
- Closed
- is related to
-
SERVER-40883 Remove the --noIndexBuildRetry server parameter flag and the storage.indexBuildRetry config file option
- Closed
- related to
-
SERVER-38351 Clear _unfinishedIndexes from IndexCatalog when restarting replica set member in standalone mode, without clearing persisted data
- Closed
-
SERVER-44206 Complete TODO listed in SERVER-37952
- Closed