grafana provides a much nicer UI than Prometheus, but Prometheus provides a more powerful alerting system featuring things like grouping/batching and silences across it’s dimensions. The alertmanager can route alerts based on labels and allows alerting on non-graphable queries. The overall pipeline also is faster since grafana needs to pull metrics from Prometheus while prometheus alerts get pushed on rule evaluation.
Therefor I’d like to keep the prometheus alerting but it would be nice if I could still edit both my dashboards and alerts in one place. Since adding such functionality is probably quite involved, I wanted to check in to see if anyone already though about this, maybe something like this is even already on the roadmap? If not, would this be something that could be accepted into Grafana or is so diametrical to the grafana alert design that it’s unlikely it would ever be added, even if we/I contribute the feature?
Hey @fish you are not the only one who finds alerts unfinished…
Truth here is that anyone will do nothing…
According to @torkel (project creator/leader) there is nothing to be done as of alerts, meaning team likes alerts in current shape and will add nothing there.
I have never said anything like that. We are very far from happy with alerting and want to way more with it, not only enhancement but rethink some of it and make new types of alert rules possible that don’t start with a dashboard panel.
But the team has been too small to handle that and simultaneously support the OSS project and enhance/modernize the core dashboarding features. We are scaling up the team quite a bit in the coming months and hope to find time to improve alerting.
Directly not, but - from reading betwen the lines - one can come to such conclusion.
Seriously?? If so, than why something that is crucial is only partly implemented?
More? Id give you advice:: stop working on Grafana, step down from being a leader and forget. Would be beneficial to all of us. Especially when you say (from other thread)::
Excuse me - if something that should be in the core from begining (proper alerting) is less important than milion other things, this clearly proves me right…
But the team has been too small to handle that and simultaneously support the OSS project and enhance/modernize the core dashboarding features. We are scaling up the team quite a bit in the coming months and hope to find time to improve alerting.
Excuse me ???Have you lost your mind??
Quoted is perfect example of underscaling. When you create project you start with as many people so they are fully capable of implementing basic core functionality/ies and - after core functionaliuty is fully implemented - you start growing your team and do other things.
=======
Not to mention deletions of unfovarable comments, and other shady stuff…
Whole Grafana thing makes me morally ashamed, I do not want OSS to look like this…
@torkel you are pretty much done here. You lost all your credibility.
what are the plans regarding alert rework? Did you consider deeper integrating with Prometheus’s alerting? No need to re-invent the wheel and would be nice to avoid competition there, given what nice fit Grafana + Prometheus are otherwise. With good alerting integration grafana could become the de facto general purpose standard frontend for Prometheus which I think would be great for both projects.
I’m happy to fill a github issue with some thoughts how this could be done.
