Proper Setup of Datadog Log Ingestion on Kubernetes
I have been working with Datadog log ingestion for about a year now. It's been (mostly) great to work with. The documentation around running it inside of Kubernetes is a bit lacking though. Their documentation covers Docker thoroughly, but Kubernetes less so.
When I installed Datadog into our Kubernetes clusters a year ago, there were two ways to do it, you could use a DaemonSet to ensure at least 1 Pod of Datadog runs on every Node. Or you could install it as a Deployment. I went with the DaemonSet option and used Helm to install it. That worked quite well!
Then we wanted to start using DogStatsD to ingest metrics about our applications, and it seemed at the time like this required the "cluster-agent" to run. I have serious doubts about this part. If I get all of the Datadog-related objects in my cluster I see the DaemonSet (daemonset.apps/dd-agent-datadog) and I also see a Deployment (daemonset..apps/dd-agent-datadog) on my cluster.
Is this right? Do I really need to run both of those things to get log ingestion and metrics?
Datadog has two agents.
- Cluster agent which is a proxy between Kubernetes API Server and Datadog node agents. The cluster agent is deployed as deployment to one of the kubernetes node.
- Node agents which is deployed in each and every Kubernetes node as Daemonset.
And yes for DogStatsD the node agents need to be deployed as Daemonset.
Here is the the deployment manifest for cluster agent and node agent.
- Automatically add imagePullSecrets to a ServiceAccount
- How can I pass environment variables to mongo docker-entrypoint-initdb.d?
- kubectl short alias for deployment
- oauth2 proxy with Ingress nginx not passing X-Auth-Request headers during standard auth flow
- Is the kubernetes kubelet a DaemonSet?
- Using /actuator/health when Spring application is deployed on k8s?
- Can you have multiple ingresses that use the same LoadBalancer?
- K3s kubeconfig authenticate with token instead of client cert
- K8S: How do I delete deployments with 0 replicas
- Kubernetes resource versioning
- How do I ssh into the VM for Minikube?
- Seed MongoDB in local minikube cluster using skaffold
- how to extract specific value from a configmap using oc client
- Question about connector plugin versions in Strimzi's Kafka Connect
- Kubectl error: memcache.go:265] couldn’t get current server API group list: Get
- Kustomize patching multiple path with same value
- Pass environment variable into a Vue App at runtime
- Unable to connect to service in same namespace in kubernetes
- Django on Kubernetes Deployment: Best practices for DB Migrations
- Helm convert data for nginx template
- kubectl error You must be logged in to the server (Unauthorized) when accessing EKS cluster
- is it possible to remote debugging java program in kubernetes using service name
- Configure Microk8s
- How to perform read queries on read replica in postgres?
- Does NodePort route traffic only to pods that are hosted under node to which we send request? How to achieve that if it's not the case?
- How do I uninstall minikube on a Mac?
- At least one invalid signature was encountered
- How to handle CPU contention for burstable k8s pods?
- Istio TLS termination and mTLS
- Helm does not upgrade rolebindings