We should allow a time-series collection to be sharded with a shard key on one of the existing indexes. Sharded time-series collections need to work with collation in the following scenarios:
- Time-series collection created with the default collation
- Sharding a time-series collection with a non-default collation.
========= OLD DESCRIPTION =======
A router (mongos) attempting an insert into a collection C, either already knows C is a time-series collection, or will receive an error return informing it of that fact. This error return must include information on what fields are the metadata and time fields. The router must be updated to check routing information for the buckets table to determine what shard to send an insert for a time-series collection to.
- is related to
-
SERVER-55426 Investigate work to support inserts on sharded time-series collections
- Closed
-
SERVER-55809 Disallow sharding a time-series collection or a buckets collection
- Closed
- related to
-
SERVER-55272 Queries on sharded time-series buckets collections are not applying shard filters
- Closed
-
SERVER-55653 Queries on time-series views do not accept collations on indexes
- Closed