All systems operational.

Last 24 hours — deploys

• Finished deploys
  289
  162 ready · 121 failed · 6 cancelled
• Success rate
  57.2%
  every finished deploy · 100.0% excluding user build errors
• Build time, median
  1 min 36 s
  p50 of the build phase
• Build time, p95
  7 min 4 s
  target under 8 min

External checks

Probed every minute from a separate machine outside the production host (hetzner-fsn1-de), through public DNS and TLS, the way a visitor reaches us. Three failed probes in a row count as an outage. The 30-day figure is what the SLA is measured against.

Service level objectives

• API availability

  met
  100.00%
  target 99.9% · 30d rolling
  one-minute probes of /health from outside the production host
• Deploy success (SLO)

  met
  100.00%
  target 99.0% · 7d rolling
  latest deploy per active project; failures caused by user code are not counted
• Build time, p95

  met
  6 min 54 s
  target under 8 min · 7d rolling
  build phase only, all finished deploys

Incidents

• Dashboard deployment pages returned 503resolvedDashboard
  2026-10-20 07:30 UTC → 2026-10-20 08:00 UTC · SEV2

  /dashboard/deployments and every deployment detail page failed to render for about 30 minutes after a web release. The API, builds, running apps and custom domains were not affected. The status page showed "All systems normal" throughout because it only probed the API; it now probes the dashboard pages directly.

  1. 07:30investigatingReports of "Couldn't load this page" on the Deployments list and on deployment detail pages after the morning web release.
  2. 07:50identifiedThe server-rendered deployments pages imported two helpers from a client-only module; Next.js rejected the call at render time, so the pages failed with a 503 while everything else was healthy.
  3. 08:00resolvedA follow-up web release moved the helpers into a shared module. Pages render again; no data was lost and no deploys were affected.
• HTTPS unavailable on buildfyio.com and all hosted appsresolvedEdge · Dashboard · API · Hosted apps
  2026-10-17 09:19 UTC → 2026-10-17 10:50 UTC · SEV1

  A configuration reload in the edge proxy crashed its TLS layer. HTTPS connections to buildfyio.com, the API and every hosted app hung for roughly 90 minutes while plain-HTTP redirects and all internal health checks kept reporting healthy. Running apps were not restarted; they were simply unreachable from the outside.

  1. 09:19investigatingHTTPS requests to the platform and to hosted apps time out. Port 80 still answers.
  2. 10:40identifiedThe edge proxy hit an internal panic during a config reload; its TLS handler and admin API hung. Internal probes never saw it because they bypass the edge.
  3. 10:50resolvedEdge proxy restarted and its routes restored. Root cause (a reconcile loop rewriting every static-site route every five minutes, ~19k reloads a day) fixed the same day; a 30-second reload grace period and a one-minute external HTTPS watchdog were added.