downscale-period not working as expected about kube-downscaler

[hjacobs]

Good point about --namespace: I added a sentence about the scenario it was designed for: c176ead

TL;DR: The usual use is to run kube-downscaler for the whole cluster and only exclude namespaces/resources.

Regarding downscale-period: what options did you specify and what other annotations did you add?

[benraven]

Thanks for the quick response, and the clarification!

That downscale-period annotation is the only one on the deployments, with no annotations on the namespace, and the only options I specified for the kube-downscaler deployment itself are --interval, --exclude-namespaces (not including the namespace that the deployments I tested are in) and --debug.

The annotation we use on everything else is downscaler/downtime: "Mon-Fri 00:00-07:00 Europe/London,Mon-Fri 19:30-24:00 Europe/London,Sat-Sun 00:00-24:00 Europe/London" which works perfectly. We use that three-part downtime instead of just downscaler/uptime: "Mon-Fri 07:00-19:30 Europe/London" because we want things that have been manually scaled down to stay scaled down until manually scaled up, and that seemed like the best way of doing that. We'd also like to be able to use downscale-period to indefinitely scale down certain things that have been manually scaled up, so that unused dev instances don't linger forever. Again, I hope I'm understanding the documentation correctly and that those are the right ways to achieve our goals.

[benraven]

#70 fixes the remaining issue I had regarding the downscaler ignoring annotations on deployments in namespaces that don't have annotations.

Unfortunately, the change to what is now Line 87 results in an error about is_uptime being referenced before declaration, but it still works.