I have a Grafana instance running against a K8s cluster and I need not as much data as what I’m currently getting by default, which is, a LOT! For example, I could restrict anything coming from the grafana or kube-system namespace (as I want to mostly focus on my own namespace and my apps there). So, ideally, I’d like to stop pushing any metrics, traces or logs from those namespaces, hence stop tracking those containers, etc.
I looked into the docs, blogs, etc. and I’m really confused. ChatGPT confused me even more… I’m reading about adaptive logs and metrics, and I’ve activated this, but this seems Grafana telling me what it thinks I should do, not what I want to do Then, there is fleet management, which is great, but it seems the self-monitoring pipelines are only for grafana itself, right? So, I’d need to create my own pipelines, but how would they relate vs. those self-monitoring (as, I suspect by removing the grafana namespace, those pipelines should not receive anything anymore, but that’s only if they run AFTER mine). Then there is the configuration of grafana per yaml through helm. Anyway, I’m really lost, any help would be appreciated, thanks in advance!
1.) Optimise/drop it at the source = drop what you don’t need at the source (K8S) level, e.g. use filtering based on namespace (of course, if you implemented fleet management, then you can use that also for this task). This is a tedious job, usually - you need to understand how everything works, how to filter, and it must be implemented in many places, …
2.) Optimise/drop it at the Grafana Cloud level - that’s exactly those “Adaptive Metrics/Logs/Traces” Grafana Cloud features. That’s simple and quick action. You can still customise what you need, usually. Grafana Cloud gives you only suggestions.
Pick&mix these options based on your needs/skills.
Jan, thank you for your reply on this. I’ll go with Adaptive Metrics/log/Traces.
But reading the documentation is not super clear (IMO), so would appreciate feedback/guidance if you have any. I created two additional segments, above and beyond the default segment that contains the SaaS’s default suggestions - which I already approved:
But that’s it, there is nothing else showing up, and the recommendation type is not something I can use to force what I’d like to see within those namespaces, it simply means (AFAIU) that it is waiting for new suggestions to come and populate that list, and the recommendation type is merely a filter to only see subsets of rules. Is my understanding correct? Is there a way to actually create a forced rule that would disregard anything coming from those ns forcefully and not wait for recommendations?
The adaptive log UI is not the same, so I’ll come back to this later as it is even more confusing to me, but I’d love to focus on metrics first.
Then, there is fleet management, which is great, but it seems the self-monitoring pipelines are only for grafana itself, right? So, I’d need to create my own pipelines, but how would they relate vs. those self-monitoring (as, I suspect by removing the grafana namespace, those pipelines should not receive anything anymore, but that’s only if they run AFTER mine). Then there is the configuration of grafana per yaml through helm. Anyway, I’m really lost, any help would be appreciated, thanks in advance!
