[ENHANCE] Cron like maintenance window for pods
Hello everyone! How do you think is it possbile to create label for allowed maintenance window for reloading pods? We have sort of applications that use certificates and after experation it should be reloaded in time for example at 2 A.M. due high load. But reloader does it straightaway configmaps/secrets changes.
Another question is how reloader detects changes that has done in time when reloader was turned off? Will it be checked by reloader or just gone?
hi @Demch1k, this feature can be added to Reloader as it aligns with the project, but we don't have active developers on this project so the community will need to work on it. PRs are welcome.
hi @MuneebAijaz , I would like to contribute as my companies platform relies on Reloader. Thus, I can help but would this be considered a good first issue for this repo? If yes, would you be open to some light shepherding?
hi @wololowarrior , I'd consider this to be a bit complex issue. I myself havent looked at the details yet of how it would be possible. Issues for newcomers are labelled as good-first-issue across the repos in opensource org. https://github.com/stakater/Reloader/issues?q=is%3Aopen+is%3Aissue+label%3Agood-first-issue
If you need this change or would still like to work on it, i can guide you
@MuneebAijaz I would like to give this a try:
- Can the maintenance window be applied at a deployment level?
- Do we want it be an annotation?
- What would its value be? a time range maybe?
- How do we handle the part where a change happened when we were not in the maintenance window?
- This mean we cannot restart the pods and any update request that comes in outside of the maintenance window might be kept somewhere to applied during the window?
Has anyone solved the maintenance window issue by scaling Reloader to 0 with a CronJob during office hours and to 1 during maintenance windows? If so, were there any issues with this approach?