Skip to content

[ML] Inference API unable to retrieve inference endpoint #132361

@jonathan-buttner

Description

@jonathan-buttner

We're seeing failures within the inference API where the internal search request is unable to retrieve the documents from the configuration and secrets indices.

Typically an error like this will be seen:

java.lang.IllegalStateException: Failed to load inference endpoint [<inference id>]. Endpoint is in an invalid state, try deleting and reinitializing the service

This could indicate that some other issue is occurring with the cluster. This is likely happening because we allow partial search results. Instead we should disable partial search results to bubble up the actual issue (whether a timeout or something else).

Metadata

Metadata

Labels

:mlMachine learning>bugTeam:MLMeta label for the ML team

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions