2026-09-04, platform
Release 1.4
Deploys roll back on a failed health check. The console gets a services table with filters.
A deploy now waits for the health check before it takes traffic. If the check fails within 30 seconds the previous version is restored and an alert goes to the service's alert email.
Changes
- The services table filters by status and by name, and remembers the selection between pages.
- Regions can be changed from the settings form. The change takes effect on the next deploy.
- Weekly digest emails are off by default for new services.
Versions
| Service | Before | After | Change |
|---|---|---|---|
| api | v2.3.8 | v2.4.1 | Health check before traffic |
| worker | v1.11.2 | v1.12.0 | Retries with backoff |
| storage | v3.0.1 | v3.0.2 | Rolled back, see the failed tab |
Scroll sideways for the rest of the table.
Configuration
Set the health check path and timeout per service.
[service.api]
health = "/healthz"
timeout = "30s"
rollback = trueNext release: per-region rollouts.