Skip to content

Preserve error that terminates a worker #4052

@coryan

Description

@coryan

The worker task for an object descriptor may terminate because the underlying connection had an unrecoverable error.

In this case, we send an error to all active ranged reads (which could be "none"). Future ranged reads get a more obscure error. We should preserve the original I/O error so we can deliver it to future reads.

That is, calling ObjectDescriptor::read_range() when the worker is terminated should return the error that terminated the worker.

Metadata

Metadata

Assignees

Labels

api: storageIssues related to the Cloud Storage API.priority: p3Desirable enhancement or fix. May not be included in next release.type: feature request‘Nice-to-have’ improvement, new feature or different behavior or design.

Type

No type

Projects

No projects

Relationships

None yet

Development

No branches or pull requests

Issue actions