Workload Orchestration

Twine

• https://engineering.fb.com/2019/06/06/data-center-engineering/twine/

Kubernetes

• https://kubernetes.io/
• https://kubernetes.io/docs/concepts/configuration/overview/#general-config-tips
• Kubernetes post-mortems

Master -> Node + Node processes

Deployment -> (service -> ) pod (on a node) -> containers/volumes

Kubectl

kubectl reference

Cluster

# get cluster info
kubectl cluster-info
# get list of nodes
kubectl get nodes

Deployments

# create a deployment
kubectl run NAME --image=...
# see deployments
kubectl get deployments

Pods

Think of a pod as a logical host

# list pods
kubectl get pods
# run a command on a pod
kubectl exec $POD_NAME COMMAND
# label a pod
kubectl label pod $POD_NAME app=v1

Networking

# Open a proxy to the internal network
kubectl proxy

kubectl expose TYPE/NAME

# Forward port 8081 to a deployment
kubectl port-forward deployment/foobar 8082

Logs

kubectl logs $POD_NAME

Services

A logical collection of pods

Type	Description
ClusterIP
NodePort
LoadBalancer	A load balancer
ExternalName

# list services
kubectl get services
# delete services
kubectl delete service

DaemonSets

Long-running processes, generally monitoring/logging. Can come up first, and go down safely at reboot/shutdown.

Namespaces

# list namespaces
kubectl get namespaces
# get detailed namespace info
kubectl describe namespace NAMESPACE
# create context from namespace
kubectl config set-context # ...
# Switch context
kubectl config use-context NAMESPACE
# Get current context
kubectl config current-context

Secrets

# Create generic secret from literal value
kubectl create secret generic OBJECTNAME --from-literal=KEYNAME=LITERAL_VALUE

https://kubernetes.io/docs/concepts/configuration/secret/

Misc

# Check control plane health
kubectl get componentstatus

minikube

https://github.com/kubernetes/minikube

Master -> Node -> Node processes

minikube start

k3

• https://github.com/rancher/k3s
• https://github.com/rancher/k3d

metrics-server

https://github.com/kubernetes-incubator/metrics-server https://kubernetes.io/docs/tasks/debug-application-cluster/core-metrics-pipeline/

heapster

https://github.com/kubernetes/heapster

deprecated Kubernetes 1.11, retired in Kubernetes 1.13

collects metrics from cadvisor

cadvisor

https://hub.docker.com/r/google/cadvisor/

collects resource/perf metrics from running containers

UI - port was 4194

in Kubernetes 1.11, UI turned off by default

Mesos

https://mesos.apache.org/

Nomad

# view registered nodes
nomad node status

# list servers
nomad server members

# Start a dev server
sudo nomad agent -dev

# run a job
nomad run JOBNAME.nomad

Default web UI port is 4646

• CLI reference

cron

• https://crontab.guru/

crontab

flag	description
-u USER	select different user
-l	display current crontab
-r	remove current crontab
-e	edit current crontab with VISUAL or EDITOR

Syntax

┌───────────── min (0 - 59)
│ ┌────────────── hour (0 - 23)
│ │ ┌─────────────── day of month (1 - 31)
│ │ │ ┌──────────────── month (1 - 12)
│ │ │ │ ┌───────────────── day of week (0 - 6) (0 to 6 are Sunday to Saturday, or use names; 7 is Sunday, the same as 0)
│ │ │ │ │
│ │ │ │ │
* * * * *  command to execute