Inhibits within before/after

[Ben Love]

I have a fairly simple setup with a handful of alerts from a specific host. If that host is down, I want to suppress these alerts. I'm able to do that with a straightforward inhibit.

- source_match:

    alertname: 'HostMissing'

  equal: ['instance']

The problem comes when the host comes back. The previously inhibited alerts continue firing for another minute or two, but the inhibit is gone, so the alerts fire notifications immediately.

Is there a way I can say "inhibit these alerts if the HostMissing alert is firing, or has been firing within the past N min"?

Or, for bonus points, is it possible to say "hold these alerts in case the inhibit begins firing within the next N min"? I realize this case would delay notifications for N min generally, but it would make nice feature parity for "within N min before".

In some sense, this is the opposite of alertmanager pull#1331. If an inhibit is defined as "within N min before", don't update the inhibitor's cache of alerts until N min have passed.