-
Type: Task
-
Resolution: Fixed
-
Priority: Major - P3
-
Affects Version/s: None
-
Component/s: None
-
None
-
Fully Compatible
-
Execution Team 2022-01-24, Execution Team 2022-02-07, Execution Team 2022-02-21, Execution Team 2022-03-07
In C2C replication, we would like to only replicate majority committed data so that we can avoid handling rollbacks of the source cluster. This is now done by using majority readConcern in collection scans in the cloning phase. However, listDatabases/listCollections/listIndexes don't support reading majority committed catalog as the catalog is not versioned. This makes it very hard for an external process to detect when the result returned by these list commands rolls back. This work is to support majority read for listDatabases by reading from a majority committed snapshot of the _mdb_catalog without actually introducing versioned catalog.
And it should also work with sharded clusters.
- is duplicated by
-
SERVER-62007 Support majority read for listCollections
- Closed
- is related to
-
SERVER-88679 Support majority read concern for listCollections, listDatabases, and listIndexes
- Backlog
- related to
-
SERVER-69750 For certain index options, $listCatalog output could have inconsistent data type as $listIndexes
- Closed
-
SERVER-62008 Support majority read for $indexStats
- Closed
-
SERVER-64251 enable feature for aggregation stage listCatalog by default
- Closed
-
SERVER-78460 Complete TODO listed in SERVER-63754
- Closed
-
SERVER-63754 avoid copying NamespaceStringOrUUID vector in OptionsWithSecondaryCollections
- Closed
-
SERVER-64203 update required permissions on listCatalog aggregation stage
- Closed
-
SERVER-64404 improve sharding support $listCatalog aggregation stage
- Closed
-
SERVER-87088 add PIT lookups to list_catalog_read_concern.js
- Closed