This page is the human-readable roll-up of the platform's
reliability monitors. It is read-only documentation of how the platform is
watched, what the current alerts are, and how to read the evidence. The live,
timestamped records live in logs/, updated by the monitoring
scripts on the owner's machine.
Status is "operational" when no monitor
is currently red. Any WARN or CRITICAL alert flips the relevant SLO below and
appends to the incident log. Full definitions in docs/slo/SLO.md.
| SLO | Status | Monitor |
|---|---|---|
| Uptime 99% | OK | tools/monitor/uptime.sh · 30s |
| Response time 1–2s | OK | tools/monitor/perf.sh · 60s |
| Booking reliability 98% | OK | tools/monitor/booking.sh |
| Data accuracy 99% | OK | tools/monitor/accuracy.sh · nightly |
| Support response 24–48h | OK | tools/monitor/support.sh |
The table reflects the most recent monitor verdict whether green or red. A red SLO flips its pill to the matching colour.
| Signal | WARN | CRITICAL | Channel(s) |
|---|---|---|---|
| Uptime | 3 fails / >1% in 5m | 10 fails / down ≥5m | email / email+WhatsApp |
| Perf p95 | >2s | >3s or >20% over 2s | email / email+WhatsApp |
| Booking | >2% fail | >5% / overbook | email / email+WhatsApp |
| Accuracy | 1 inaccuracy | >1% / cross-academy | email / email+WhatsApp |
| Support | >24h | >48h | email / email+WhatsApp |
WhatsApp carries CRITICAL only, and its text is sanitised to
never name a child, academy, email, phone, or request body (the lock-screen
rule). The durable incident logbook is docs/pages/reliability.md.
logs/uptime.log — one row per probe (code, latency, ok).logs/perf.log — millisecond latencies for p95.logs/booking.log, logs/support.log, logs/accuracy.log — counts.docs/pages/reliability.md — the incident log appended by _alert.sh.Suggested fixes for any red SLO are recorded in the incident
log as each alert fires; the human process (escalation) is in
docs/slo/SLO.md.