Skip to content

Remote bridge maps terminal store failures to transient offline errors #86

Description

@danny-avila

Summary

RemoteBridgeSandboxBackend.execute() handles a few BridgeStoreError codes explicitly, then maps every remaining code to BRIDGE_WORKER_OFFLINE with transient=true.

This includes terminal conditions such as:

  • WORKSPACE_QUARANTINED
  • WORKER_MISMATCH
  • WORKER_QUARANTINED
  • ASSIGNMENT_FENCED

These conditions are not temporary worker outages, and retrying with fresh identifiers does not repair them.

Impact

Callers can retry an operation that should remain fenced or rejected, while the surfaced BRIDGE_WORKER_OFFLINE code hides the actual recovery action. In particular, a quarantined workspace requires an explicit reset rather than an automatic retry.

Suggested direction

Handle terminal bridge-store codes explicitly and return a non-transient backend error. An exhaustive mapping (or an explicit default for truly availability-related failures only) would prevent new store codes from silently becoming retryable.

Add focused tests covering mismatch, quarantine, and fenced errors and asserting both the public backend error code and transient === false.

Verified on current main (0eb0f3a30e984f23fa98d0cc48acc8abd21cb184).

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions