Skip to content

Troubleshooting

Nick Hamze edited this page Sep 4, 2026 · 4 revisions

Troubleshooting

Old UI

Verify Fleet's installed version, active plugin folder, and actual asset hashes. Reload the shell and clear relevant caches. Deterministic ZIP timestamps can confuse local OPcache when repeatedly extracting in place; restart that fixture's PHP or touch its deployed files.

Framework missing

Fleet has no classic fallback. Install a verified App Framework build. A matching version string alone is not enough during experimental development.

Password or connection problems

Sign-in happens on the managed WordPress site, not in Fleet. Review the domain and approve the separate Application Password.

Check public HTTPS, the full subdirectory address, Core's /?rest_route=/ index, Application Password support, firewalls, and Authorization-header forwarding. Private production addresses are intentionally blocked.

Start a fresh check after expiry and remain the same signed-in hub user. Do not share old callback URLs. Revoke an unused remote Fleet credential if verification failed after approval.

Setup fails

Approval is retained. Fix filesystem permissions, interactive FTP requirements, blocked WordPress.org access, administrator capabilities, or OpenStation requirements; retry Finish setup. Manual OpenStation installation is another option.

401, 403, or offline

Use Repair connection for a revoked credential. A 403 may be a destination permission/security policy. Offline sites back off while other sites continue. Manual checks bypass the wait.

Status checks verify authenticated administrator access, not just the public API index. Failed verification preserves the last cached inbox and enters backoff.

Repair keeps agency details. An old-password cleanup warning requires checking the managed user's Application Password list and revoking the obsolete entry.

Local Studio health warnings

The demo server can report a ten-second loopback timeout when Core's Site Health test runs through authenticated REST, even when its WP-CLI loopback succeeds. Fleet keeps that reported warning visible; a green connection badge means authentication works, not that every Site Health probe passed. Check the same workflow on the production hosting matrix before treating local results as a hosting diagnosis.

Conflicts and uncertain saves

Copy retained edits, reopen the current version, and reconcile. No automatic merge or remote lock is promised. After uncertain draft creation, search WordPress before making another. Unsaved source is not a durable autosave.

Missing/stale results

Apply collection filters again or return to page one. Fleet-wide Search reads a periodic encrypted index; Inbox/health are cached too. Quiet hubs need server-driven WP-Cron. Explorer cannot invent a missing endpoint or upload binary files.

Disconnect and support

A revocation failure retains the connection for repair/retry. Disconnect before uninstalling.

Copy the redacted report from Support and describe steps/expected/actual results in a GitHub issue. Exclude callback URLs, credentials, headers, salts, private content, and authorization screenshots.

Clone this wiki locally