Status
Status
Checks this Worker ran on the request that rendered this page. It is not independently monitored: this page is served by the same Worker it reports on, so an outage that takes the Worker down takes this page with it.
| Component | State | Detail |
|---|---|---|
| Database (D1) SELECT 1 through this Worker's own D1 binding | reachable | SELECT 1 round trip in 1 ms |
| Object store (R2) HEAD of a probe key through this Worker's own R2 binding | reachable | bucket answered a HEAD in 1 ms |
| Delivery of this page none — see the detail | self-report | this page was served, so the Worker and the edge in front of it ran for THIS request. It is circular: had either been down you would have received an error page from Cloudflare, not this row saying so. |
| Enrolled machines none from here — see the detail | not measured here | each enrolled machine is reachable only with its own per-box credential, which this public page must not hold. Signed in, the Machines page shows per-box reachability. |
Checks above ran when you loaded this page: 2026-10-01T18:38:08.556Z. Reload to run them again; nothing here is cached or scheduled.
Running build: 57784ea22f3823c9c801851140cddf8daf4a0b4f (57784ea2)
What this page does not know
It cannot report its own outage. Every probe above runs inside the Worker. If the Worker, its edge, or DNS is down, you do not see a red row — you see an error page from Cloudflare, or nothing. A status page that answers only while the service is healthy cannot tell you the service is unhealthy, and this one does not pretend it can. Satisfying that properly needs a probe running on infrastructure Hawkeye does not operate, on a schedule, and that does not exist yet.
There is no incident history here. No incident record is published, so this page shows none. That is an absence of records, not a record of no incidents — do not read the empty space as a clean history, and do not read it as uptime. No uptime percentage is published because none has been measured over any defined window.
A grey pill is not a green one. not measured here means nothing probed that component on this request. It is not a claim that the component is healthy, and it never turns green on its own — the rows that say it have no probe wired to them at all.
TODO — not filled in yet
These are deliberately blank rather than invented. Nothing below is a placeholder standing in for a real answer; where a fact is missing, it is missing.
- No independent monitoring. A real status page needs probes from at least two regions on a third-party platform (Better Stack, Checkly or equivalent), with incident states, subscriber notifications and on-call paging. None of that is provisioned: it needs an external account and credentials, which is not something this repository can create for itself. Until it exists, this page states only what the Worker can observe about itself and says so at the top.
- No component history and no incident timeline. Nothing records these readings anywhere, so there is no 30-day component history, no incident state machine, no postmortem link and no maintenance window. Each page load starts from nothing.
- No per-machine status. Enrolled machines are reachable only with their own per-box credentials, which a public page must not hold. The Machines page, signed in, is where per-box reachability lives.
Machine-readable: GET /healthz returns
{"ok":true,"service":"hawkeye-panel"} when the Worker itself is
serving requests. It touches neither D1 nor R2, so it answers a narrower
question than this page does.