Vikrant Singh

← How this site is run

Reliability

This site runs one small scheduled job and is checked from outside every 5 minutes. Below is how that is going, straight from its own records.

All checks normal · no open incidents · dependencies OK · data since 2026-10-18

Right now

Operating normally

Last check
9 min ago
Last shipped
7 days ago (1610101)
Build
success

Availability (outside probe)

Black-box SLI · good = GET /api/pulse answers 200 within 2 s · every 5 min

100%

8,442 of 8,442 checks good

Objective: 99% over 30 days

Error budget: 0 of 86 allowed failures used · 86 left

An uptime monitor outside Cloudflare asks this site for its status snapshot. The site answers 200 only while the snapshot is under 35 minutes old, so one check covers both: the site is up, and its data is fresh.

Cron job success (GitHub sync)

White-box SLI · good = the scheduled run completes · every 10 min

99.95%

4,281 of 4,283 checks good

Objective: 98% over 30 days

Error budget: 2 of 86 allowed failures used · 84 left

A cron-triggered Worker reads this site's latest commit and CI result from the GitHub API and saves them as the status snapshot. This is the site's own record of the job the probe depends on: when it keeps failing, the probe's figure follows.

Daily record, last 30 days

Availability (outside probe)

  1. 2026-10-18: 283 of 283 good
  2. 2026-10-19: 284 of 284 good
  3. 2026-10-20: 285 of 285 good
  4. 2026-10-21: 284 of 284 good
  5. 2026-10-22: 283 of 283 good
  6. 2026-10-23: 282 of 282 good
  7. 2026-10-24: 285 of 285 good
  8. 2026-10-25: 283 of 283 good
  9. 2026-10-26: 283 of 283 good
  10. 2026-10-27: 284 of 284 good
  11. 2026-10-28: 285 of 285 good
  12. 2026-10-29: 291 of 291 good
  13. 2026-10-30: 284 of 284 good
  14. 2026-10-31: 283 of 283 good
  15. 2026-11-01: 296 of 296 good
  16. 2026-11-02: 284 of 284 good
  17. 2026-11-03: 283 of 283 good
  18. 2026-11-04: 286 of 286 good
  19. 2026-11-05: 283 of 283 good
  20. 2026-11-06: 284 of 284 good
  21. 2026-11-07: 283 of 283 good
  22. 2026-11-08: 285 of 285 good
  23. 2026-11-09: 283 of 283 good
  24. 2026-11-10: 284 of 284 good
  25. 2026-11-11: 283 of 283 good
  26. 2026-11-12: 284 of 284 good
  27. 2026-11-13: 283 of 283 good
  28. 2026-11-14: 284 of 284 good
  29. 2026-11-15: 283 of 283 good
  30. 2026-11-16: 195 of 195 good

Cron job success (GitHub sync)

  1. 2026-10-18: 143 of 144 good
  2. 2026-10-19: 144 of 144 good
  3. 2026-10-20: 144 of 144 good
  4. 2026-10-21: 144 of 144 good
  5. 2026-10-22: 144 of 144 good
  6. 2026-10-23: 144 of 144 good
  7. 2026-10-24: 144 of 144 good
  8. 2026-10-25: 144 of 144 good
  9. 2026-10-26: 144 of 144 good
  10. 2026-10-27: 143 of 144 good
  11. 2026-10-28: 145 of 145 good
  12. 2026-10-29: 145 of 145 good
  13. 2026-10-30: 144 of 144 good
  14. 2026-10-31: 144 of 144 good
  15. 2026-11-01: 150 of 150 good
  16. 2026-11-02: 144 of 144 good
  17. 2026-11-03: 144 of 144 good
  18. 2026-11-04: 144 of 144 good
  19. 2026-11-05: 144 of 144 good
  20. 2026-11-06: 144 of 144 good
  21. 2026-11-07: 144 of 144 good
  22. 2026-11-08: 144 of 144 good
  23. 2026-11-09: 144 of 144 good
  24. 2026-11-10: 144 of 144 good
  25. 2026-11-11: 144 of 144 good
  26. 2026-11-12: 144 of 144 good
  27. 2026-11-13: 144 of 144 good
  28. 2026-11-14: 144 of 144 good
  29. 2026-11-15: 144 of 144 good
  30. 2026-11-16: 99 of 99 good

2026-10-18today

no failures some failures, objective met objective missed that day no data

Latency by day (p95) · 95% of checks finished within this time

Times are rounded up to the nearest band (50, 100, 200, 400, 800, 1,600, 3,200 ms).

DayCron job (GitHub API read)Outside probe (/api/pulse answer)
2026-11-16≤ 1,600 ms≤ 400 ms
2026-11-15≤ 1,600 ms≤ 200 ms
2026-11-14≤ 1,600 ms≤ 200 ms
2026-11-13≤ 1,600 ms≤ 200 ms
2026-11-12≤ 1,600 ms≤ 200 ms
2026-11-11≤ 1,600 ms≤ 200 ms
2026-11-10≤ 1,600 ms≤ 200 ms

Failures

  1. Cron job failed: TimeoutError (cause: github)
  2. Cron job failed: TimeoutError (cause: github)

Full failure history →

37 slow but successful runs (over half the time budget) also recorded as early warnings.

What could stop this site

Everything this site depends on that expires, and what breaks if it does. Dates are read every day from the service that issued them; one that can't be read is shown as a risk, not as fine. Renewal reminders go to the owner at 90, 60 and 30 days.

  • OKDomain vikrantsingh.fyi29 Sep 2027 · 317 days

    domain registry, checked 17 h ago

  • OKGitHub token (scheduled job)30 Aug 2027 · 286 days

    read from GitHub, checked 17 h ago

  • OKIncident desk token30 Sep 2027 · 317 days

    read from GitHub, checked 17 h ago

  • OKLog backfill token31 Oct 2027 · 349 days

    read from Cloudflare, checked 17 h ago

  • OKUsage feed token30 Nov 2027 · 379 days

    read from Cloudflare, checked 17 h ago

  • OKCloudflare deploy token31 Mar 2028 · 501 days

    reported by the deploy pipeline, checked 26 h ago

How this board found its own blind spot →

Today's free allowance

  • Database rows written: 577 of 100,000 (1%)

  • Database rows read: 197,433 of 5,000,000 (4%)

  • Database requests: 761 of 100,000 (1%)

  • Site requests: 1,613 of 100,000 (2%)

  • Status store writes: 312 of 1,000 (31%)

Yesterday (UTC day 2026-11-15): database rows written 724 (1%) · database rows read 222,190 (4%) · database requests 929 (1%) · site requests 4,559 (5%) · status store writes 361 (36%).

As of 16:20 EST; resets at 19:00 EST (00:00 UTC). Shared by every Worker on the account. Running out pauses the live numbers and Ludo until the reset, and says so; it is never an incident. Total storage isn't shown: Cloudflare's analytics give no storage figure for this account.

Incident desk

Real failures of this site go to an incident triage agent. It proposes priority and drafts updates; a human runs every step after that, in public GitHub issues.

  1. CLOSEDLOW vikrantsingh.fyi heartbeat: Scheduled job failed 2 runs in a row: injected 503 (github) (game day: github_5xx)

    Opened 2026-10-02 00:50 EDT · AI proposed Low (confidence 0.92) · gate: approved by human · 2 drafts, never sent · site recovered 2026-10-02 01:10 EDT

    site raised it → AI proposed → human: ACKNOWLEDGED → ACTIVE → RESOLVED → CLOSED

2 earlier incidents
  1. CLOSEDLOW manual test: Grok key test: scheduled job failed 2 runs in a row: http 503 (github) (not a real outage)

    Opened 2026-10-01 01:04 EDT · AI proposed Low (confidence 0.85) · gate: auto-applied under the earlier low-risk rule (before 1 Oct 2026); AI priority now waits for /approve · 2 drafts, never sent · site recovered 2026-10-01 01:40 EDT

    Owner's note: Game day 1 (planned fault, FAULT=github_5xx): the site's real alert landed on this test ticket (finding 1, since fixed). See the postmortem: /notes/postmortem-game-day-1/

    site raised it → AI proposed → human: ACKNOWLEDGED → ACTIVE → RESOLVED → CLOSED

  2. CLOSED manual test: bridge smoke test (not a real outage)

    Opened 2026-10-01 00:30 EDT · No AI proposal (model unavailable) · gate: model unavailable; human triage

    site raised it → AI unavailable → human: ACKNOWLEDGED → ACTIVE → RESOLVED → CLOSED

Low-priority tickets

Below incident level: they never page anyone and are never counted as incidents.

Renewal reminders

Raised by the daily dependency check: a renewal or expiry the owner needs to act on before anything breaks. They open at P3 (90 days out), escalate to P2, P1 and P0 as the date nears, and close themselves once fixed.

No open tickets.

6 closed tickets
  1. CLOSEDP3Domain vikrantsingh.fyi #8

    Opened 2026-11-07 13:03 EST · closed 2026-11-08 13:22 EST

  2. CLOSEDP3Domain vikrantsingh.fyi #7

    Opened 2026-10-28 14:52 EDT · closed 2026-10-29 14:58 EDT

  3. CLOSEDP3Domain vikrantsingh.fyi #5

    Opened 2026-10-12 17:45 EDT · closed 2026-10-13 15:52 EDT

  4. CLOSEDP3Domain vikrantsingh.fyi #4

    Opened 2026-10-08 15:17 EDT · closed 2026-10-09 14:49 EDT

  5. CLOSEDP3Usage feed token #3

    Opened 2026-10-04 18:28 EDT · closed 2026-10-04 19:43 EDT

  6. CLOSEDP3Domain vikrantsingh.fyi #2

    Opened 2026-10-04 18:28 EDT · closed 2026-10-04 19:43 EDT

As of the daily check, 2026-11-15 14:23 EST · the dependency board

Failed workflow runs · last 30 days

The build-and-deploy pipeline and the daily dependency check run as GitHub Actions. When one fails, the live site keeps serving the last good version, so it is listed here rather than raised as an incident.

None in the last 30 days.

Read hourly from GitHub; as of 2026-11-16 16:00 EST · all runs

Incidents are alerts from this site only; tests are titled as tests. AI drafts are never sent. Since 1 Oct 2026, every AI priority waits for a human approval; each entry's gate says which rule applied when it was triaged. AI priority never pages anyone: paging comes from the outside probe on this site's SLO. After intake, every state change is a human command. Engine repo →

Client path: browser telemetry (Pulse run)

Real-user monitoring (RUM): a 30-second game times itself in the visitor's browser and sends its errors and frame times in one beacon. Reported by the browser, not measured by the server.

Sessions
4 (4 played)
Session error rate
0%
sessions with a JavaScript error
Frame time (p95)
≤ 50 ms
60 fps needs about 16 ms
Jank rate
0%
sessions with a frame over 50 ms

Add a data point: play Pulse run for 30 seconds →

Client-only signal. Does not prove server capacity or ITSM readiness.

Server path: WebSocket game server (Ludo)

Every roll and move is decided and timed inside the game room on the server (a Durable Object), so these are measured, not reported by browsers. Objectives are proposed and not yet held for a full 30 days.

Connection success
100% (objective 99.50%)
WebSocket join answered: 101, or a correct refusal such as “full” · 9 checks · p95 ≤ 800 ms
Action latency
100% (objective 99.00%)
receive → save → broadcast ≤ 100 ms · 469 checks · p95 ≤ 50 ms
Bot turn lag
100% (objective 99.00%)
each bot step ≤ 250 ms past its pace · 17 checks · p95 ≤ 50 ms
Round-trip time
66.03% (objective 95.00%)
server ping → page echo ≤ 300 ms · 368 checks · p95 ≤ 800 ms
Engagement · completion, lobby starts, turn timeouts
Game completion
50.00%
1 of 2 reached a winner
Lobby start rate
25.00%
within 2 minutes · 4 code rooms
Turn timeout rate
12.35%
30 s without a move · median think ≤ 3,200 ms today

Add a data point: play Ludo → · raw data

How to read this
  • Objective (SLO): the share of checks that must succeed over 30 days.
  • Error budget: how many failures the objective allows in 30 days. Spending it is fine; overspending means reliability work comes before new features.
  • The official external number is the outside monitor's 30-day uptime, reported monthly. The figures here are the site's own records of the same checks.
  • All times and days are Toronto time (ET: EDT in summer, EST in winter).

Page requests served in this window: 3,993 (bots included) · served without the live status line: 0.

Updated 2026-11-16 16:29 EST · failure history · raw data (JSON) · dependencies (JSON)