Site Uptime
The rolling-window uptime percentage for each monitored CanadaLogin site, showing what proportion of five-minute pings succeeded over each time horizon.
Six sites are monitored: the two public-facing login pages (EN/FR), the Manage Profile app (staging and production), and the IdP (staging and production).
Calculation
Filter to the most recent date partition and read the precomputed rolling-window uptime columns from sre.uptime_summary. Each uptime_* column is already a percentage (0-100), calculated from all five-minute pings over the named window.
To track uptime over time for a specific site, remove the MAX(date) filter and group by date.
Upstream monitoring gap (July 28 – September 23, 2026): External synthetic health checks in cds-snc/status-statut halted during this ~8-week window. Because zero down minutes were recorded across zero executed pings, sre.uptime_summary defaulted to reporting uptime_day = 100.0% with response_day_ms = 0.
- When aggregating daily uptime or querying historical trends, filter out unmonitored days using
WHERE response_day_ms > 0. - In snapshot queries, treat
response_day_ms = 0(orNULL) as unmonitored rather than confirmed 100% availability.
Latest snapshot, all sites, all rolling windows:
SELECT
slug,
lang,
status,
uptime_day,
uptime_week,
uptime_month,
uptime_year,
uptime_alltime
FROM sre.uptime_summary
WHERE date = (SELECT MAX(date) FROM sre.uptime_summary)
ORDER BY slug;Trend over time, year uptime for a specific site:
SELECT
date,
uptime_year,
uptime_alltime
FROM sre.uptime_summary
-- set the reporting window; see the timespans note in cookbook/notes.qmd
WHERE slug = 'canada-login-website-en'
AND CAST(date AS DATE) BETWEEN DATE '2026-01-01' AND DATE '2026-06-30'
-- exclude dates where upstream monitoring was inactive
AND response_day_ms > 0
ORDER BY date;Latest snapshot, all sites, all rolling windows:
uptime_s <- tbl(con, in_schema("sre", "uptime_summary"))
latest_date <- uptime_s |> summarise(d = max(date, na.rm = TRUE)) |> pull(d)
uptime_s |>
filter(as.Date(date) == as.Date(latest_date)) |>
select(slug, lang, status, uptime_day, uptime_week, uptime_month,
uptime_year, uptime_alltime) |>
arrange(slug)Trend over time, year uptime for a specific site:
tbl(con, in_schema("sre", "uptime_summary")) |>
# set the reporting window; see the timespans note in cookbook/notes.qmd
filter(
slug == "canada-login-website-en",
as.Date(date) >= as.Date("2026-01-01"),
as.Date(date) <= as.Date("2026-06-30"),
# exclude dates where upstream monitoring was inactive
response_day_ms > 0
) |>
select(date, uptime_year, uptime_alltime) |>
arrange(date)Interpretation notes
The uptime windows are rolling, not calendar-aligned. uptime_month on a given date covers the 30 days trailing that date’s stats update (~23:2x UTC), not a fixed calendar month. To aggregate by calendar month, group by substr(date, 1, 7).
Rolling windows (uptime_week, uptime_month, uptime_year, uptime_alltime) spanning the July 28 – September 23, 2026 gap incorporate unmonitored days recorded as 100% uptime, which may slightly elevate rolling availability figures.
Alternative names
- uptime percentage
- availability percentage
- site availability
- service uptime by site