Sometimes our users have the need to ignore some of our KubeOptimizer recommendations on certain workloads and keep their original assigned request and limit settings, and mostly for good reasons. You might have a workload that you are ok with having it throttling cause they are a low priority for your business, or you have a workload with very unpredictable behavior and you want to assign high requests and limits to avoid throttling or OOM kills.
In these cases, the recommendations you don't want is just noise and create confusion for you and your team. That's why we decided to introduce this feature, which enables you to have an annotation in your workload specs to instruct Magalix KubeOptimizer to bypass this workload and ignore generating any recommendations for it.
How it works
It's really simple to use this feature, by just adding the following annotation to your workloads specs, you can control whether you get KubeOptimizer recommendations or not.
magalix.com/kube-optimizer : “false | true”
By default, KubeOptimizer will generate recommendations for all your workloads. If you set the above annotation to true the KubeOptimizer will generate recommendations, if you set it to false KubeOptimizer will ignore that workload and you won't see any optimization recommendations showing up in your Magalix dashboards.