Go concurrency · Cloudflare edge

Whimsy's Warden

Keeping watch over everything Whimsy runs.

Whimsy Warden is an availability and latency monitor for production applications and stable beta deployments. It combines a concurrent Go probe engine with scheduled edge checks, a responsive status UI, persistent state, and actionable e-mail alerts.

Live Whimsy Warden dashboard showing operational services, latency, e-mail alert policy, and production and beta systems
Service health Monitor active
Operational9
Degraded0
Median latency184 ms
@Service alerts are armedThreshold 500 ms · duplicates suppressed
Whimsy productionLast edge check just nowHealthy
Stable betaIndependent environmentHealthy
Live operational dashboard Open Warden
GoBounded concurrent probes
5 minutesScheduled edge checks
500 msLatency alert threshold
KVSnapshots and alert state

Fast checks without uncontrolled concurrency.

The Go monitor runs target probes through a bounded worker pool. Context cancellation and explicit timeouts prevent a slow service from holding the run open, while connection reuse, redirect limits, and modern TLS defaults keep the HTTP client efficient and deliberate.

A shared declarative target inventory covers both production and stable beta applications. Results are returned in deterministic order, which keeps command-line output, dashboard data, and tests predictable even though the network work happens concurrently.

Probe, persist, decide, notify.

Schedule

A Cloudflare Worker cron starts a complete check every five minutes.

Probe

Targets are checked for availability and response latency.

Persist

The latest snapshot and incident state are written to Cloudflare KV.

Alert

Cloudflare Email Service reports new outages and latency breaches as they occur.

E-mail for transitions, not noise.

Warden alerts the Whimsy monitoring inbox when a service goes down or crosses the 500 ms latency threshold. Incident state in KV suppresses duplicates until recovery, so a continuing failure does not create a new message every five minutes.

A control room that reads at a glance.

Status at a glance.

The dependency-free dashboard surfaces healthy and degraded counts, median latency, last-check time, and individual targets. Production and beta filters reduce clutter, while a manual refresh can request a new edge check when an operator needs current evidence. Responsive layouts reorganize the summary, filters, and service cards across wide desktop, tablet, and mobile resolutions so the same operational picture remains useful on a workstation or on the go.

One system, safely separated environments.

One system, safely separated environments.

The UI and API are served at the Cloudflare edge with no-store responses and structured observability logs. Production and beta deploy independently; beta has its own KV namespace and no alerting cron, preventing test traffic from duplicating production incidents.

GoCloudflare WorkersWorkers KVEmail Service
TypeScriptAccessible UIResponsive design
GoCloudflare WorkersWorkers KV
Email ServiceTypeScript
Accessible UIResponsive design