Maintenance Schedules
Organisation-wide maintenance windows that decide when AME upgrades clusters and reboots nodes
A maintenance schedule defines the windows in which AME may perform disruptive maintenance on a cluster. Schedules belong to your organisation, so you define a window once and assign it to as many clusters as you like.
A cluster uses a maintenance schedule for two kinds of maintenance:
- Automatic cluster upgrades: AME starts an auto upgrade only inside the window of the cluster's upgrade schedule. The window decides when an upgrade starts, not when it ends: AME does not pause or stop a running upgrade when the window closes.
- System reboots: AME reboots nodes that require a reboot after OS updates only inside the window of the cluster's reboot schedule. See automatic node reboots.
A cluster can use the same schedule for both, or a separate schedule for each.
You manage schedules on the Schedules tab of the Maintenance page in the Avisi Cloud Console. The overview lists each schedule with its window, its time zone and the number of clusters that use it.
To create, edit or assign a schedule, see managing maintenance schedules.
Windows
A schedule consists of one or more windows. Each window has:
- Days: the days of the week on which a window starts.
- Start time and end time: when the window opens and closes. An end time before the start time means the window ends on the next day, so a window from 22:00 to 04:00 runs overnight.
- Time zone: the time zone in which the start and end times apply. AME applies daylight saving time automatically, so a window from 00:00 to 06:00 in
Europe/Amsterdamkeeps those local times all year round.
With Same window on every day enabled, every selected day uses the same start and end time. Turn it off to give each day its own window.
Schedules for system reboots
Not every schedule can be used for system reboots. A reboot schedule must meet two requirements:
- it uses the same window on every day it covers, and
- its window ends on the day it starts, so it does not run past midnight.
A schedule such as "every day, 00:00 to 06:00" or "Tuesday to Sunday, 00:00 to 04:00" qualifies. A schedule with a different window on Saturday, or a window from 22:00 to 04:00, can only be used for upgrades.
The console only offers qualifying schedules when you select a reboot schedule. When you edit a schedule that clusters use for system reboots, the editor shows which clusters use it and reminds you of these requirements.
Default schedules
Your organisation has a default schedule for upgrades and a default schedule for reboots. AME assigns these defaults to every new cluster, and you can change the schedule of each cluster afterwards. Changing a default does not affect existing clusters.
Only a schedule that meets the reboot requirements can be the default for reboots.
Overlapping upgrade and reboot windows
Both cluster upgrades and system reboots drain nodes. When the upgrade schedule and the reboot schedule of a cluster overlap, an upgrade and a reboot can drain nodes at the same time, which reduces the capacity available to your workloads. The cluster's Patching page warns you when the two schedules overlap.
Keep upgrade and reboot windows apart
Give a cluster an upgrade schedule and a reboot schedule that do not overlap, for example upgrades on Sunday morning and reboots on weekday nights.
Unused schedules
The overview shows how many clusters use each schedule. Delete unused schedules removes every schedule that no cluster uses, which keeps the list short as clusters come and go.