Skip to content

Handover timeout can cause namespace to be stuck in unavailable state #8229

@meiliang86

Description

@meiliang86

If the handover workflow is invoked with a small timeout, it is possible that the execution times out between the two namespaces updates, thus could leave the namespace in an unavailable state.

The workaround is to always involke the handover workflow with a timeout >> 1hr (the max timeout for catchup activity). So we could add validations in the beginning of the workflow to fail it immediately if the workflow execution timeout is too small.

We should improve the handover workflow to backout before changing the ns to be unavailable if it's too close to execution timeout.

Metadata

Metadata

Assignees

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions