Skip to content

[subnet] Retry evaluation runs under specific conditions #418

Description

@jmnmv12

Whenever a transient failure on an evaluation run happens the platform needs to analyse and retry the evaluation run if possible. We need to spike to understand in what conditions the retry makes sense. E.g. If retrying the eval run can put the agent on top it make sense to retry.

Right now an evaluation run can only be executed once. We need to expand this concept to allow for retries, capturing the results of each individual retry. We also need to explore ways of triggering this retry, platform retry vs validator retry.

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Fields

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