If uptime and downtime overlap the replicaset will be scaled down

[fbrnc]

kube-downscaler/kube_downscaler/scaler.py

Line 53 in 1d5da01

is_uptime = helper.matches_time_spec(now, uptime) and not helper.matches_time_spec(now, downtime)

In case uptime and downtime overlap this line will evaluate to is_uptime = false resulting in the replica sets to be scaled down.
I wonder if this is excepted/wanted behavior. In case of doubt (e.g. conflicing configuration) I'd rather keep a replica set up and running instead of shutting it down :)

[hjacobs]

Yes, this behavior is expected, but we can obviously discuss what makes sense in general.

[fbrnc]

Maybe we could introduce an Apache-style (order deny,allow) flag that controls if uptime or downtime has a higher priority for overlapping configurations, defaulting to downtime having a higher priority for backwards-compatibility.