Skip to main content

Outage Events

Individual site outages: when a site went down, when it recovered, and the status code.

Individual site outage events detected by the uptime monitor: when a site went down, when it recovered, and the HTTP status code at time of detection.

Calculation

Filter sre.uptime_timeseries to rows where status = 'down'. Each such row represents a moment when the upptime bot detected the site was unreachable and committed that state to the history file.

To count outage events per site over a reporting period, group the filtered rows by slug and count.

All outage events in a reporting period:

SELECT
    date,
    ping_timestamp,
    slug,
    lang,
    http_code,
    response_ms
FROM sre.uptime_timeseries
-- set the reporting window; see the timespans note in cookbook/notes.qmd
WHERE status = 'down'
  AND CAST(date AS DATE) BETWEEN DATE '2026-01-01' AND DATE '2026-06-30'
ORDER BY date DESC, slug;

Outage event count per site in a reporting period:

SELECT
    slug,
    lang,
    COUNT(*) AS outage_event_count
FROM sre.uptime_timeseries
-- set the reporting window; see the timespans note in cookbook/notes.qmd
WHERE status = 'down'
  AND CAST(date AS DATE) BETWEEN DATE '2026-01-01' AND DATE '2026-06-30'
GROUP BY slug, lang
ORDER BY outage_event_count DESC;

All outage events in a reporting period:

tbl(con, in_schema("sre", "uptime_timeseries")) |>
  # set the reporting window; see the timespans note in cookbook/notes.qmd
  filter(
    status == "down",
    as.Date(date) >= as.Date("2026-01-01"),
    as.Date(date) <= as.Date("2026-06-30")
  ) |>
  select(date, ping_timestamp, slug, lang, http_code, response_ms) |>
  arrange(desc(date), slug)

Outage event count per site in a reporting period:

tbl(con, in_schema("sre", "uptime_timeseries")) |>
  # set the reporting window; see the timespans note in cookbook/notes.qmd
  filter(
    status == "down",
    as.Date(date) >= as.Date("2026-01-01"),
    as.Date(date) <= as.Date("2026-06-30")
  ) |>
  count(slug, lang, name = "outage_event_count") |>
  arrange(desc(outage_event_count))

Interpretation notes

Each row in sre.uptime_timeseries is a single commit by the upptime bot. On a normal all-up day there is one row per site (the daily ~23:2x UTC snapshot). On an outage day there are additional rows (one when the site went down and one when it recovered), so status = 'down' rows count detection events, not total minutes of downtime. For total outage duration, use the daily_minutes_down JSON field in sre.uptime_summary.

During the July 28 – September 23, 2026 upstream monitoring outage, zero ping commits were executed. The absence of status = 'down' rows during that window reflects inactive synthetic monitoring rather than confirmed 100% uptime.

Alternative names

  • downtime events
  • site outages
  • availability incidents
  • uptime incidents