Repository navigation

keep-alive #1246

	name: keep-alive

	# Resurrection knock for the Render free-tier dyno. The app keeps ITSELF awake
	# once running (self-ping thread in cloud/app.py — see "self keep-alive"); this
	# workflow just wakes it back up if it ever sleeps (quiet-hours window, Render
	# maintenance restart). GitHub dequeues */5 crons every 1.5-2.5h in practice,
	# which is fine for that job but useless as the primary keep-warm — that's why
	# the self-ping exists. Auto-disables after 60 days of no repo commits.

	on:
	schedule:
	# Every 5 min, but only ~7am-1am Central (12:00-06:00 UTC) so we don't burn
	# the whole 750 free instance-hours/month keeping it awake at 4am.
	- cron: "/5 0-6,12-23 * *"
	workflow_dispatch: {}

	jobs:
	ping:
	runs-on: ubuntu-latest
	steps:
	- name: ping render /healthz
	run: \|
	URL="https://transcript-app-cloud.onrender.com/healthz"
	for i in 1 2 3; do
	code=$(curl -s -o /dev/null -w "%{http_code}" --max-time 60 "$URL" \|\| true)
	echo "attempt $i -> HTTP $code"
	[ "$code" = "200" ] && exit 0
	sleep 20
	done
	echo "healthz never returned 200"; exit 1