error playground

every button is a real production failure mode → a real Prometheus alert → a real Discord ping.
pod: …
active storms: none

🔥 error rate

3 minutes of 5xx responses on /api/work under synthetic load (~10 rps).

fires PlaygroundHighErrorRate in ~2 min

🐌 latency

4 minutes of 1.5–3.5s responses. p99 goes through the roof.

fires PlaygroundHighLatencyP99 in ~2-3 min

💀 crash

Process exits 1. Kubernetes restarts the pod — press 2-3× over a couple minutes to build a restart streak.

fires PlaygroundCrashLooping after ~2 restarts

🎈 memory leak

Balloons memory ~8Mi/50ms past the 128Mi container limit until the kernel OOM-kills the pod.

fires PlaygroundOOMKilled within ~2 min

⚡ cpu burn

Spins every core for 4 minutes against a 250m cpu limit — heavy cgroup throttling.

fires PlaygroundCPUThrottling in ~3 min

🪵 log spam

Floods ERROR-level structured logs for 60s (100/s). Feeds log alerting once Loki lands.

visible in kubectl logs now; Loki rule later
crash + memory leak kill the pod — the dashboard will blip while kubernetes restarts it (that's the point). metrics at /metrics · health at /health