-
Notifications
You must be signed in to change notification settings - Fork 0
Release Notes
Kody Jordan edited this page Aug 13, 2026
·
17 revisions
Version 1.6.0 makes the application degrade gracefully when the PostgREST API host is unreachable. Previously an API outage left users on a blank page for the full network timeout, and sign-in failed with a raw Failed to fetch error. The app now paints immediately, bounds every API request with a timeout, and reports outages in plain language.
This release includes:
- Added a shared request timeout to every PostgREST call so an unreachable host fails in seconds instead of stalling on the browser network timeout.
- Removed runtime configuration loading from the critical render path so the sign-in page appears immediately.
- Added an API availability store that converts network failures into a readable service-unavailable message.
- Added a static outage switch that can publish a maintenance message while the database is unavailable.
- Added a service notice banner and connection retry to the sign-in page.
QualityChecks v1.6.0 replaces blank screens and raw fetch errors during an API outage with fast, readable feedback, and adds a maintenance announcement path that does not depend on the database.
- All PostgREST reads, writes, RPC calls, and health checks share a single timeout wrapper with an 8-second default.
- Each call site can override the timeout; startup configuration uses a shorter 4-second bound.
- Caller-initiated cancellations, such as navigating away from a page, are no longer misreported as service outages.
- The application shell mounts before runtime configuration resolves.
- Application settings and role values load concurrently rather than sequentially.
- Built-in defaults cover authentication and role configuration when settings cannot be loaded.
- Network failures now surface a service-unavailable message instead of a raw fetch error.
- The sign-in page shows a service notice banner at the top of the page and a connection retry action within the sign-in card.
- Microsoft sign-in is disabled while the service is unreachable, so users are not sent through a redirect that cannot succeed.
- Signed-in sessions show a service banner when the connection is lost mid-session.
- A deployed
status.jsonfile can publish a maintenance message when the database-backed settings are unreachable. - Maintenance messages support links so users can be pointed to an external status page.
- Static maintenance notices are advisory only and do not redirect users to the maintenance page, so sign-in resumes as soon as service is restored.
- Database-driven maintenance mode continues to redirect non-admin users as before.
- Added tests for request timeout behavior and abort handling.
- Added tests for availability state transitions on network failure and recovery.
- Fixed a blank page on load when the API host was unreachable.
- Fixed sign-in hanging for the full network timeout before failing.
- Fixed raw
Failed to fetchtext appearing as the sign-in error message. - Fixed maintenance mode being unavailable as an announcement channel during a database outage.
- No database schema migration required.
- No Power Automate changes.
-
src/public/status.jsonships with maintenance disabled by default. Enable it only to announce an outage, and reset it once service is restored.