For existing sharded time-series collections, it is possible to have unset fields like granularity or bucketMaxSpanSeconds in the sharding catalog. Some callers that read these fields expect valid values, so we should either generate them on the fly while reading them, or fix them in the sharding catalog lazily.
- is related to
-
SERVER-76411 Set missing time-series options fields in sharding catalog during upgrade
- Backlog
-
SERVER-76242 Write a customized IDL parser for time-series options
- Backlog