Refactor some ILM tests to not use deprecated methods #131417
Conversation
And updates `MetadataMigrateToDataTiersRoutingService` to accept a `ProjectState` instead of a `ProjectResolver`.
nielsbauman
added
>test
elasticsearchmachine
added
Team:Data Management
|
Pinging @elastic/es-data-management (Team:Data Management) |
| ); | ||
| return Tuple.tuple( | ||
| ClusterState.builder(currentState).metadata(mb).putProjectMetadata(newProjectMetadataBuilder).build(), | ||
| ClusterState.builder(currentState.cluster()).metadata(mb).putProjectMetadata(newProjectMetadataBuilder).build(), |
There was a problem hiding this comment.
Your change here seems like a straight refactoring, so that's fine, but I have a wider question. This code is working at project scope, but it's also mutating state at the cluster scope (removing the ENFORCE_DEFAULT_TIER_PREFERENCE setting). That can't be safe, can it? Do we just not care because it's ILM? If so, it feels like we should use your annotation to call that out. If we plan to make it safe, is that tracked somewhere?
There was a problem hiding this comment.
Yeah I agree that it's not ideal, but I figured we can make an exception for ILM. I have a feeling that we'll be making some changes to ProjectState at some point, to avoid these kinds of things (i.e. to make it safer in terms of what is modifiable). When we do so, we'll have to take places like this into account. Adding a @NotMultiProjectCapable sounds like a good idea. Will do.
|
Nit: I just noticed you applied the |
nielsbauman
added
>non-issue
and removed
>test
| ); | ||
| return Tuple.tuple( | ||
| ClusterState.builder(currentState).metadata(mb).putProjectMetadata(newProjectMetadataBuilder).build(), | ||
| ClusterState.builder(currentState.cluster()).metadata(mb).putProjectMetadata(newProjectMetadataBuilder).build(), |
|
@elasticmachine update branch |
4 participants
And updates
MetadataMigrateToDataTiersRoutingServiceto accept aProjectStateinstead of aProjectResolver.