2026.10.1.03 — Startup recovery for managed macOS workers

All release notes

2026-10-01 · Creator Publishing Hub managed macOS worker support

This update adds startup recovery for required worker storage and an existing authenticated remote connection, with a 15-minute storage recovery window and protection against repeated automatic restarts during the same outage.

What changed

• Check whether the configured storage volume is available after startup and attempt to mount it when the expected volume is detected.
• Allow a 15-minute recovery window before requesting one orderly automatic restart if the required storage remains unavailable.
• Cancel the pending restart when storage becomes healthy and retain outage state across startup to prevent repeated automatic restarts during the same unresolved outage.
• Resume the existing authenticated worker connection after startup when its saved identity matches the configured worker, without enrolling a new device or changing its connection settings.
• Provide a read-only worker storage-status command that can report the countdown without first opening the external home directory.

Verification

• 49 deterministic storage-watchdog tests and 29 connection-helper tests passed, including recovery cancellation, cross-boot reboot-budget persistence, exact volume and connection identity checks, bounded probes, and failure handling.
• Shell validation and nine isolated status-command checks passed. The fixed status command was also verified through the collaborator workspace using its actual SSH connection.
• Both background services are installed in the system startup domain. Their first live runs exited successfully and reported healthy state; repeated storage checks were also observed. Installed artifact hashes match the reviewed sources.
• The collaborator account passed actual nested-directory creation, a disposable Python environment, Git initialization, compiler execution, and SFTP upload/download checks after its access correction.
• An automatic missing-storage reboot and the new connection helper running after a fresh boot have not been physically exercised. Their failure and recovery decisions were tested with simulated inputs and bounded runtime checks.

Rollout

The storage component has been superseded by version 2026.10.1.04 following its managed deployment. The private-network connection helper remains at version 2026.10.1.03. The installation and healthy-state checks recorded for this release remain historical verification; an automatic missing-storage reboot and connection recovery after a fresh boot have not been physically exercised.

Updating

• This service is installed and configured by the managed worker administrator. Website owners do not need to update WordPress or change publishing settings.
• Preserve the recovery state when updating the service. Rollback disables the service and restores its preceding configuration without removing user files.

Scope and limitations

• The watchdog cannot repair a damaged drive, cable, filesystem, or unavailable storage device.
• A continuing outage receives at most one automatic restart. Further recovery requires operator attention until the volume has been confirmed healthy.
• This release does not establish that audio generation, publishing, or social delivery is operational. Those workflows require their own verification.
• The connection helper depends on a previously authenticated connection profile and the supported local networking service. It does not sign in or replace a missing profile.
• These services check periodically, so scheduling and bounded probe time can extend the recovery window beyond 15 minutes. They do not intentionally reboot before the full window has elapsed.
• The initial storage countdown used a process-relative clock on the worker Python runtime, so elapsed time was not reliably carried across separate service runs. The storage component in version 2026.10.1.04 corrects this limitation. The version 2026.10.1.03 private-network connection helper is unaffected.

Contact CPH support