Declare planned downtime
Who acts: any person on the team, in MMHotline's pages; or the monitoring tool itself, when it lists its own server's maintenance.
When to use this page: you are about to stop a monitoring tool that checks in with MMHotline, such as to move or upgrade its server, and nobody should be woken because its check-ins stop.
What this page does: you tell MMHotline until when one tool's silence is expected, so it sends no alert about missing check-ins until then.
Terms:
- A check-in is the signed message a monitoring tool sends every 30 seconds to say it still runs. When check-ins stop, MMHotline sends a Warning and then an SOS, as Connect a tool that sends alerts sets up.
- Planned downtime stops those alerts about missing check-ins for one tool, until a chosen time. Alerts the tool still sends arrive as usual.
Choose the route
- You stop the tool, or its server, by hand: Start planned downtime in MMHotline.
- The tool's own server goes into maintenance, and the tool knows it: Let the tool report its maintenance.
- You want no phone to ring for a while, for any reason: an administrator mutes the team, as Mute the team says.
Start planned downtime
-
Open the tool's page. In MMHotline, open Settings → Alert sources, then the tool, such as Monitoring. The page is Check-ins and planned downtime.
-
Fill in Planned downtime. Set Until to the date and time the tool should be back, in the team's time zone named beside the field, at most 7 days ahead. It starts one hour from now; for 30 minutes, change the time to 30 minutes from now. Optionally enter Reason (optional), such as
moving the monitoring server, at most 200 characters. Choose Start planned downtime.Check: the panel says:
Planned downtime Until 02:45, in 59 minutes, started by Carlos Example: moving the monitoring server. End planned downtime nowThe alert history records it as an Info alert, which wakes nobody:
Carlos Example started planned downtime for alert source “Monitoring” until 2026-11-15 07:45:00 UTCsudo mmhotline alert-sources listalso shows the downtime:Planned downtime until 2026-11-15T07:45:00Z, started by Carlos Example: moving the monitoring server. -
Do the work. While the downtime lasts, the tool's silence sends nothing.
-
When the tool is back early, choose End planned downtime now. Otherwise the downtime ends by itself at its end time.
Check: the history records
Carlos Example ended planned downtime for alert source “Monitoring”, and the page's Last check-in from ... line shows the next check-ins.
Downtime always ends, so a forgotten downtime can't silence a tool for good. While it lasts, the page offers only the end button; to change the end time, end it and start a new one. A tool switched off in the settings file can't get downtime; the page asks you to enable it first.
After the downtime ends
If the tool is still silent, MMHotline sends an SOS, with no Warning first,
once both of these have passed: one minute after the downtime ended, and
the tool's usual SOS time since it was last heard from. The tool's alerts
count as hearing from it too, but only a check-in clears an alert about
missing check-ins. The Warning and SOS times are the tool's
warn_after_no_check_in_for_minutes and
sos_after_no_check_in_for_minutes in the settings file, as
Connect a tool that sends alerts
says. Reason: you expected
the tool back by then, so its silence now is a real problem.
Let the tool report its maintenance
A tool can list servers in maintenance in each check-in, in
nodes_in_whole_node_maintenance. When that list holds the server the
check-ins come from, sent_by_node_name, MMHotline counts it as planned
downtime until the window's maintenance_until:
{"sent_at":"2026-11-15T06:50:00Z","sent_by_node_name":"monitor01","nodes_in_whole_node_maintenance":[{"node_name":"monitor01","maintenance_until":"2026-11-15T08:00:00Z","reason":"moving the monitoring server","started_by":"barbara"}]}
- A window for other servers, or for only some checks, does not count: those servers' own problems still reach MMHotline as alerts.
- The page names the tool's window; end it in the tool, not in MMHotline.
- If the check-ins stop, MMHotline keeps the last window it received until that window's end, but at most 7 days after the check-in that listed it. Reason: a window must always end. While check-ins keep listing the window, MMHotline moves that limit on.
If something is different
- The page says
planned downtime must end in the future, at most 7 days from now: choose an end time in that range. Reason: downtime must always end. - The page says
the planned downtime reason must be at most 200 characters: shorten the reason. - The tool's ordinary alerts still arrive during downtime: planned downtime stops only alerts about missing check-ins. To stop phones ringing, an administrator mutes the team.
- An SOS about missing check-ins came after the downtime: the tool did not come back in time. It resolves by itself when check-ins return, unless someone claimed it; then the person who claimed it resolves it.
This page is docs/how-to/declare-planned-downtime.md in the MMHotline repository.