What the new dashboard, Services page and Hosts page need from obs-beacond:
each attention item names its open incident (id, opened, who acknowledged it), so the page can offer "I'm on it"
each person's previous visit (a visit ends after 30 minutes without the page open), and the incidents opened or resolved since then, with how many paged someone. No previous visit: the last 12 hours
each service's last 24 hours as 48 half-hour cells (the worst status in each) from status_log, and its uptime over 30 days
a watch list in place of the gauges: disks filling up (the line fit reports already use), hosts busier than usual for the hour (now against the same hour over the past week), and what changed in the last day (reboots, new hosts)
The slow parts are worked out at most once a minute, not on every poll: at thousands of hosts the overview is read every few seconds by every open dashboard.
What the new dashboard, Services page and Hosts page need from obs-beacond:
- each attention item names its open incident (id, opened, who acknowledged it), so the page can offer "I'm on it"
- each person's previous visit (a visit ends after 30 minutes without the page open), and the incidents opened or resolved since then, with how many paged someone. No previous visit: the last 12 hours
- each service's last 24 hours as 48 half-hour cells (the worst status in each) from status_log, and its uptime over 30 days
- a watch list in place of the gauges: disks filling up (the line fit reports already use), hosts busier than usual for the hour (now against the same hour over the past week), and what changed in the last day (reboots, new hosts)
The slow parts are worked out at most once a minute, not on every poll: at thousands of hosts the overview is read every few seconds by every open dashboard.
Done in #73. The overview now carries today (incidents opened and closed since your previous visit, which ends after 30 minutes away, or the last 12 hours with none on record or for demo visitors), services (each service's last 24 hours as 48 half-hour cells plus 24 h and 30 d uptime; GET /api/v1/services adds parts, needs and needed-by), watch (disks filling within 30 days, hosts busier than usual for the hour, hosts restarted in the last day), and each attention item's open incident with who acknowledged it. History is read at most once a minute (30 days and the watch list every ten). Checked against the demo with 30 days of history; documented in docs/formats.md, The dashboard's briefing.
Done in #73. The overview now carries `today` (incidents opened and closed since your previous visit, which ends after 30 minutes away, or the last 12 hours with none on record or for demo visitors), `services` (each service's last 24 hours as 48 half-hour cells plus 24 h and 30 d uptime; `GET /api/v1/services` adds parts, needs and needed-by), `watch` (disks filling within 30 days, hosts busier than usual for the hour, hosts restarted in the last day), and each attention item's open `incident` with who acknowledged it. History is read at most once a minute (30 days and the watch list every ten). Checked against the demo with 30 days of history; documented in docs/formats.md, The dashboard's briefing.
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
What the new dashboard, Services page and Hosts page need from obs-beacond:
The slow parts are worked out at most once a minute, not on every poll: at thousands of hosts the overview is read every few seconds by every open dashboard.
Done in #73. The overview now carries
today(incidents opened and closed since your previous visit, which ends after 30 minutes away, or the last 12 hours with none on record or for demo visitors),services(each service's last 24 hours as 48 half-hour cells plus 24 h and 30 d uptime;GET /api/v1/servicesadds parts, needs and needed-by),watch(disks filling within 30 days, hosts busier than usual for the hour, hosts restarted in the last day), and each attention item's openincidentwith who acknowledged it. History is read at most once a minute (30 days and the watch list every ten). Checked against the demo with 30 days of history; documented in docs/formats.md, The dashboard's briefing.