I've been wrestling with a frustrating issue involving ArgoCD and Crossplane, where ArgoCD shows my resources as "Healthy" and "Synced" despite Crossplane failing to provision AWS resources. I'm seeing 400 errors from AWS, but ArgoCD is blissfully unaware, maintaining its cool facade. This leads to problems like Lambda functions not updating, stuck RDS instances, and IAM roles not being created. It seems like no one's talking about this issue online, and I'm starting to think I'm the only one experiencing this strange silence on health checks. Has anyone else run into this? Are people just ignoring health checks and monitoring AWS directly? Did I stumble onto some rare configuration no one else touches? I ended up fixing the logic to prioritize error conditions, but it's baffling that this isn't more widely acknowledged. If I'm pioneering something here, I just want to know I'm not alone!
5 Answers
Glad you found a workaround! But, dude, Medium as a member-only article? Not cool! Sharing info should be accessible to everyone. Maybe consider just posting the solution on GitHub instead?
Totally! Medium can be a pain, especially when it's behind a paywall.
Thanks for your insights! We're considering switching to the same tech stack you're using, and your experiences might save us some frustration down the line. Have you thought about filing a GitHub issue for this? It might help others too!
Seems like a wise option, but if the maintainers aren’t prioritizing it, I guess a Medium article was the next best thing.
Why does this feel like it's more suited for GitHub than Medium? If it's truly a problem, open an issue there instead of making it a blog!
Right? The maintainers don’t consider it a pressing issue apparently.
I've dealt with this issue a while back, fortunately knowing how Argo health checks work helped me create my own tests early on. It seems like a lot of folks just need to custom code their health checks as using defaults isn't cutting it.
You're spot on! A lot of people aren't aware of these nuances. Custom checks are a must.
You've got it a bit wrong, mate. ArgoCD is doing its job—the resources are synced as they should be! Just because Crossplane encounters errors after that doesn't mean Argo's incorrect. GitOps is about ensuring your cluster matches your declared state; real-time health monitoring is a separate concern altogether. You might want to look into actual alerting tools for that.
Exactly! It's better to think of ArgoCD as a deployment tool rather than your overall health dashboard. Look into Grafana or Prometheus for deeper health checks.
Yeah, this is pretty much correct. ArgoCD's alerts mainly show that resources are synced or in a crash loop, but it won’t catch every hiccup unless you have proper monitoring in place.
I avoid Medium altogether. If it's member-only, I'm out.