-
Notifications
You must be signed in to change notification settings - Fork 2.8k
Closed
Labels
c/serverRelated to serverRelated to servere/intermediatecan be wrapped up in a weekcan be wrapped up in a weekk/enhancementNew feature or improve an existing featureNew feature or improve an existing featurep/highcandidate for being included in the upcoming sprintcandidate for being included in the upcoming sprint
Description
The reload_metadata API does not reload Remote Schemas as it is a costly step (and might be unnecessary). For reloading remote schemas, there is the explicit reload_remote_schema API.
But there might be a need to reload_metadata while also reloading Remote Schemas. Hence, an optional argument could be provided to reload_metadata like reload_remote_schemas: true to achieve this.
Side note: While writing down this issue, I noticed a small inconsistency in the behaviour (before v1.1):
- Suppose you reload_metadata or reload_remote_schema in one Hasura instance, this changed the schema cache and triggered a postgres event to refresh other instances.
- The other instances build schema cache from scratch and hence have up-to-date schema. Whereas the originator instance may not have up-to-date remote schema or up-to-date postgres schema (as it only did a partial reload).
@lexi-lambda Is this issue moot after v1.1 because of incremental schema cache building?
Metadata
Metadata
Assignees
Labels
c/serverRelated to serverRelated to servere/intermediatecan be wrapped up in a weekcan be wrapped up in a weekk/enhancementNew feature or improve an existing featureNew feature or improve an existing featurep/highcandidate for being included in the upcoming sprintcandidate for being included in the upcoming sprint