Add Kubernetes(Collector) #5
Conversation
|
Thank you for starting this. |
collector/kubernetes.go
Outdated
| var wg sync.WaitGroup | ||
| for _, nodepool := range cluster.NodePools { | ||
| wg.Add(1) | ||
| go func(np *godo.KubernetesNodePool) { |
There was a problem hiding this comment.
We don't really need this goroutine here, do we? It should just work without it and then we don't need the working group either.
Am I missing something?
There was a problem hiding this comment.
I was anticipating having more work to do calculating metrics for the NodePools; I'll remove.
There was a problem hiding this comment.
Resolved: 4b6ac4c
|
I think this is a suitable basic implementation. |
|
Alright. One last comment about the metric names. - digitalocean_kubernetes_cluster_count 1
+ digitalocean_kubernetes_clusters 1
digitalocean_kubernetes_cluster_up{id="f74f45ec-0c53-4e96-8f22-XXX",name="metalmatze",region="fra1",version="1.14.8-do.2"} 1
- digitalocean_kubernetes_nodepools_count{id="f74f45ec-0c53-4e96-8f22-XXX",name="metalmatze",region="fra1",version="1.14.8-do.2"} 1
+ digitalocean_kubernetes_nodepools{id="f74f45ec-0c53-4e96-8f22-XXX",name="metalmatze",region="fra1",version="1.14.8-do.2"} 1
- digitalocean_kubernetes_nodes_count{id="5879b3c5-1b25-49d3-9ff3-XXX",name="pool-0dgh51ahd",region="fra1"} 3
+ digitalocean_kubernetes_nodes{id="5879b3c5-1b25-49d3-9ff3-XXX",name="pool-0dgh51ahd",region="fra1"} 3If you want to, you can rebase against the new current master to make use of Go modules and the activated Drone CI, otherwise, we can simply merge it too, as I verified it locally. 😊 |
|
I realized my mistake in computing I will remove the metric entirely. Count should be computed by summing the e.g. |
|
Ready when you are. Is this good to go from your side? |
|
Thanks a lot for this valuable contribution! 🎉 |
What do you think to my progressing this?
It provides basic (cluster, nodepool, node) counts for Digital Ocean's Kubernetes service.
digitalocean_errors_total{collector="kubernetes"} 0 # HELP digitalocean_kubernetes_cluster_count Number of Kubernetes clusters # TYPE digitalocean_kubernetes_cluster_count gauge digitalocean_kubernetes_cluster_count 1 # HELP digitalocean_kubernetes_cluster_up If 1 the kubernetes cluster is up and running, 0 otherwise # TYPE digitalocean_kubernetes_cluster_up gauge digitalocean_kubernetes_cluster_up{id="[[REDACTED]]",name="[[REDACTED]]",region="sfo2"} 0 # HELP digitalocean_kubernetes_nodepools_count Number of Kubernetes nodepools # TYPE digitalocean_kubernetes_nodepools_count gauge digitalocean_kubernetes_nodepools_count{id="[[REDACTED]]",name="[[REDACTED]]",region="sfo2"} 1 # HELP digitalocean_kubernetes_nodes_count Number of Kubernetes nodes # TYPE digitalocean_kubernetes_nodes_count gauge digitalocean_kubernetes_nodes_count{id="[[REDACTED]]",name="[[REDACTED]]",region=""} 3By the way, this inspired a similar solution for Linode (link). My motivation is to standardize on a way to monitor and alert on resources I'm using across multiple cloud platforms (Google Cloud Platform, Digital Ocean and Linode) in order to keep my bills reasonable.
Since you created this, Go Modules are now a good package management solution for Golang.If you're supportive, I propose replacing dep with Go Modules?