Skip to content

[ML] Return 408 instead of 500 when open/start requests time out #89585

@dolaru

Description

@dolaru

Description

This applies to the following APIs:

  • open anomaly detection
  • start datafeed
  • start data frame analytics job
  • start trained model deployment

Currently, if the operation takes longer than the configured timeout, the response status code is 500.

With Kibana in mind, a 4xx response would be more suitable as it will lead to showing more useful information to the user about why the request failed.

Looking at RFC 9110, it seems the most appropriate 4xx code would be 408 Request Timeout, even though the RFC describes a different situation.

The alternative, 504 Gateway Timeout, would be more appropriate for this situation but it wouldn't improve the user experience in Kibana.

Metadata

Metadata

Labels

Type

No type
No fields configured for issues without a type.

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions