type: decision
status: active
timestamp: 2026-06-22
tags: [status, monitoring, uptime, cloudflare, workers, kv, telegram, rss, decisions, architecture]

oriz-status-app — self-hosted status page replaces UptimeRobot + Better Stack

Locked 2026-06-22. In-house status page at status.oriz.in CF Worker cron every 5 min probes every URL in FAMILY_* registries, writes to KV, served by sibling read-only Worker behind 60-sec edge cache. Replaces UptimeRobot (commercial-use ban Oct 2024) and supersedes the 10-monitor Better Stack ceiling. Telegram alerts on transition. RSS feed for incidents. 30/90-day uptime rollups.'

oriz-status-app — self-hosted status page

Decision

The oriz.in family’s status surface is now a self-hosted app at status.oriz.in, backed by two Cloudflare Workers and one KV namespace. UptimeRobot is dropped (commercial-use clause added Oct 2024 — incompatible with the family’s free-tier rule). Better Stack remains optional as a “second opinion” probe during the transition but is no longer the source of truth.

Architecture

                       +- status.oriz.in (CF Pages, Astro static)
                       ¦      ¦ fetch /api/status every 60 s
                       ¦      ?
  +- oriz-status-ping ---? STATUS_KV ?-- oriz-status-api ?- status-api.oriz.in
  ¦   (cron */5 min)      (KV namespace)   (read-only)
  ¦                          ¦
  ¦                          +-- latest          (current snapshot, no TTL)
  ¦                          +-- previous        (snapshot from prior tick)
  ¦                          +-- history:YYYY-MM-DD (per-day rollup, 90d TTL)
  ¦                          +-- incidents        (last 50 transitions, capped)
  ¦
  +-- on transition ? Telegram bot (TELEGRAM_BOT_TOKEN + TELEGRAM_OPS_CHAT_ID)

Frontend — Astro static, deployed to CF Pages. SSR placeholders for every target; client-side JS hydrates the dots from the API.
oriz-status-ping — cron Worker, */5 * * * *. Probes every target in workers/targets.ts (mirrors FAMILY_* from @chirag127/astro-shell). HEAD-first with GET fallback on 405/501. Treats >3000 ms as degraded. Writes latest, rolls history: for the day, appends to incidents on transitions, fires Telegram.
oriz-status-api — read-only Worker. Serves /api/status, /api/uptime?slug=&days=, /api/incidents, /feed.xml. 60-sec Cache-Control and matching s-maxage for edge caching.

Free-tier headroom math

Quota	Limit	Our usage	Headroom
Workers Free invocations	100,000 / day	288 cron + ~1,440 API reads = ~1,728	~58×
Workers Free outbound subrequests	50 per invocation	24 targets, batched 25 at a time	OK
KV read	100,000 / day	~1,440 API reads × 1-2 KV reads = ~2,880	~35×
KV write	1,000 / day	288 ticks × ~4 writes = ~1,150	tight — see note
KV storage	1 GB	~10 KB per day rollup × 90 = ~900 KB	massive

KV write headroom note: 1,152 writes/day uses 115% of the 1,000/day free quota. Mitigation: batch the four writes into one latest blob that embeds previous, incidents, and the per-day rollup keyed inside the same JSON — collapses to ~288 writes/day. Implement before we cross the limit (we have a 4-day window of headroom). Logged as a TODO in oriz-status-app#issues.

Telegram alerts

On every status transition (up ? degraded ? down), the cron Worker POSTs a Markdown message to the ops chat. Token + chat-id are set via wrangler secret put on oriz-status-ping; both are optional — if missing the alert no-ops. The same Telegram bot serves the wider auto-tracking pipeline.

RSS feed

status-api.oriz.in/feed.xml emits an RFC RSS 2.0 document built from the incidents KV blob (last 50 transitions). /feed.xml on the Pages domain redirects to the canonical worker route via public/_redirects.

Why not UptimeRobot

UptimeRobot revised its TOS October 2024 to prohibit use by any entity engaging in commercial activity, “even tangentially”. The oriz.in family runs ads, affiliate links, and Razorpay payments in parallel; tangential commercial activity is undeniable. Tool exited.

Why not Better Stack alone

Better Stack’s free tier is capped at 10 monitors. The family already exceeds 24 targets (live) and is growing toward 50+ as the APIs spin up. Capping monitoring at 10 means flying blind on ~80% of surfaces. Self-hosting on CF Free has comfortable headroom to 50+ targets without paying.

Surfaces probed (as of 2026-06-22)

1 master apex (oriz.in)
18 app subdomains (one-level only, per Rule 16)
5 live API subdomains (-api form, also one-level per Rule 16)

Future additions: as *-api.oriz.in and book-*.oriz.in go live, append to workers/targets.ts and src/data/targets.ts (until the codegen script that pulls from @chirag127/astro-shell lands).

The existing per-site <StatusBanner /> (see status-banner-on-every-site) switches its feed source from Better Stack’s RSS to https://status-api.oriz.in/feed.xml. One-line change in @chirag127/oriz-ui once the worker is live and a few incidents have published.

Edit on GitHub · Back to index