Skip to main content

sre.uptime_summary

Daily rolling-window uptime and response-time figures for each monitored site.

One row per monitored site per day. Each row is a snapshot of rolling-window statistics fetched from history/summary.json in the cds-snc/status-statut GitHub repository at 04:00 ET.

Provenance

Fully automated. No manual upload required. The upptime summary.yml GitHub Action recalculates all rolling-window metrics from the full five-minute ping history once per day (~00:00 UTC), rewriting summary.json. The Lambda runs at 04:00 ET and fetches that file, writing one Parquet row per site into the date=YYYY-MM-DD partition corresponding to the preceding day.

  • Updated: Daily, at ~04:00 ET
  • Retention: 7 years
  • Steward: EDCP Data and Research Team
  • Contact: Yinlin Zhang (Data and Research team)

Six monitored sites:

slug lang url
canada-login-website-en WEB-EN https://login.canada.ca
canada-login-website-fr WEB-FR https://connexion.canada.ca
canada-login-manage-profile-staging MP-STAGING https://app.login-connexion.alpha.canada.ca
canada-login-manage-profile-prod MP-PROD https://app.login-connexion.canada.ca
canada-login-idp-staging IDP-STAGING https://auth.login-connexion.alpha.canada.ca
canada-login-idp-prod IDP-PROD https://auth.login-connexion.canada.ca

Fields

Field Name Type Description
date varchar Hive partition key: the date the Lambda ran for (yesterday’s date, YYYY-MM-DD).
fetch_date varchar UTC timestamp when the Lambda fetched the data (YYYY-MM-DD HH:MM:SS). Useful for diagnosing staleness.
slug varchar Upptime site identifier. One of the six values in the table above.
lang varchar Short label identifying the site variant, one of the six values in the table above.
url varchar The monitored URL.
status varchar Site status at fetch time: up or down.
uptime_day double Uptime percentage over the past 24 hours (~288 five-minute pings).
uptime_week double Uptime percentage over the past 7 days (~2,016 pings).
uptime_month double Uptime percentage over the past 30 days (~8,640 pings).
uptime_year double Uptime percentage over the past 365 days (~105,120 pings).
uptime_alltime double Uptime percentage across all pings since monitoring began (April 2, 2026).
response_day_ms bigint Average response time over the past 24 hours, in milliseconds. Equals 0 during unmonitored outage gaps (see notes).
response_week_ms bigint Average response time over the past 7 days, in milliseconds.
response_month_ms bigint Average response time over the past 30 days, in milliseconds.
response_year_ms bigint Average response time over the past 365 days, in milliseconds.
response_alltime_ms bigint Average response time across all pings since monitoring began, in milliseconds.
daily_minutes_down varchar JSON string mapping each outage date to total minutes down on that date.

Notes

Timing note. When the Lambda runs on June 30, it fetches rolling-window statistics last recalculated at ~23:2x UTC on June 29. The resulting row lands in partition date=2026-06-29. All rolling windows trail backward from that reference point.

Upstream monitoring gap (July 28 – September 23, 2026). Automated synthetic probes in cds-snc/status-statut halted during this period due to GitHub Actions workflow execution protections. Daily summary rows were still ingested with uptime_day = 100.0, but with response_day_ms = 0 serving as a natural sentinel indicator. Exclude these unmonitored dates in queries with WHERE response_day_ms > 0.