test(infra): verify-observability asserts Grafana + Tempo/Prometheus datasources are live (refs #122)

Polls Grafana's datasource-health endpoints so it proves Grafana can actually
reach Tempo and Prometheus over the cg network, not merely that the containers
started. Fails until the backplane is wired (next commit).

refs #122
This commit is contained in:
not
2026-07-23 12:41:58 +02:00
parent 4fe9915816
commit f12e9ce746
2 changed files with 41 additions and 1 deletions
+35
View File
@@ -0,0 +1,35 @@
#!/usr/bin/env bash
#
# S-16a (#122): assert the observability backplane is live against the running stack.
# Not just "the containers started" — it asks Grafana to health-check its provisioned
# Tempo and Prometheus datasources, which proves Grafana can actually reach both over
# the `cg` network and that provisioning landed. Polls, so it tolerates a cold Grafana.
#
# Usage: run-observability-check.sh (override GRAFANA_URL / GRAFANA_AUTH / OBS_TIMEOUT)
set -euo pipefail
GRAFANA="${GRAFANA_URL:-http://localhost:3000}"
AUTH="${GRAFANA_AUTH:-admin:admin}"
TIMEOUT="${OBS_TIMEOUT:-60}"
# poll <description> <url> <json-path-expr> <expected>
poll() {
local desc="$1" url="$2" path="$3" want="$4" got deadline
deadline=$(( $(date +%s) + TIMEOUT ))
while :; do
got="$(curl -fsS -u "$AUTH" "$url" 2>/dev/null \
| python3 -c "import sys,json;print(json.load(sys.stdin)$path)" 2>/dev/null || true)"
[ "$got" = "$want" ] && { echo "$desc"; return 0; }
if [ "$(date +%s)" -ge "$deadline" ]; then
echo "$desc — got '$got', want '$want' ($url)" >&2
return 1
fi
sleep 2
done
}
echo "Checking observability backplane at $GRAFANA ..."
poll "Grafana is healthy" "$GRAFANA/api/health" "['database']" "ok"
poll "Prometheus datasource reachable" "$GRAFANA/api/datasources/uid/prometheus/health" "['status']" "OK"
poll "Tempo datasource reachable" "$GRAFANA/api/datasources/uid/tempo/health" "['status']" "OK"
echo "Observability backplane OK."