59 lines
2.3 KiB
Markdown
59 lines
2.3 KiB
Markdown
|
#k8x - deploys k8s with 3 masters
|
||
|
|
|
||
|
|
tested OS:
|
||
|
|
- debian 10
|
||
|
|
|
||
|
|
# Configuration
|
||
|
|
|
||
|
|
config.dist - copy to config and setup the cluster deployment specific settings.
|
||
|
|
|
||
|
|
config-coreapps.dist - copy to config-coreapps and setup the core applications specific settings.
|
||
|
|
|
||
|
|
# Installation of cluster nodes
|
||
|
|
|
||
|
|
./build_crypto.sh - run once to generate a CA and pack it as cryptopack.b64 file which should be shared on all nodes
|
||
|
|
|
||
|
|
./init.sh - install k8s master or worker capabilities on each kubernetes node. requires the initial hostname to be set up as those listed in the config file. run simultaniously on all masters to setup etcd, and then once on each worker node.
|
||
|
|
|
||
|
|
# Installation of core applications
|
||
|
|
|
||
|
|
Note: Run the scripts from a node which has a config-coreapps config file
|
||
|
|
|
||
|
|
./deploy_system.sh - installs the calico sdn, coredns and metrics server into the kube-system namespace, setups helm and kubelet rbac
|
||
|
|
|
||
|
|
./deploy_cert_manager.sh - installs the letsencrypt acme ssl certificate manager into the cert-manager namespace
|
||
|
|
|
||
|
|
./deploy_ingress_nginx.sh - install the ingress-nginx ingress controller into the ingress-nginx namespace
|
||
|
|
|
||
|
|
./attach_storage_ceph.sh - create ceph storage keys into kube-system namespace and the rados-block storage class which pvcs could attach to for dynamic provisioning
|
||
|
|
|
||
|
|
./deploy_monitoring.sh - install prometheus, loki, grafana and zabbix
|
||
|
|
|
||
|
|
# Installation of additional environments
|
||
|
|
|
||
|
|
./deploy_env.sh staging - install the staging namespace with wildcard ssl
|
||
|
|
|
||
|
|
./deploy_env.sh develop - install the develop namespace with wildcard ssl, and common docker registry
|
||
|
|
|
||
|
|
./deploy_env.sh live - install the production namespace
|
||
|
|
|
||
|
|
# Tools
|
||
|
|
|
||
|
|
./health.sh - displays the cluster health
|
||
|
|
|
||
|
|
./health.sh objects - displays the cluster and cluster objects health
|
||
|
|
|
||
|
|
k9s - start the kubernetes ncurses admin panel
|
||
|
|
|
||
|
|
watch -n 5 "k get pods --sort-by=.status.startTime --all-namespaces -o wide | tac" - watch for cluster pod state changes
|
||
|
|
|
||
|
|
source config ; export KUBECONFIG=$ADMIN_KUBECONFIG - exports the cluster admin key to use the kubectl (or k) command cluster-wide
|
||
|
|
|
||
|
|
# Additional Tasks
|
||
|
|
|
||
|
|
1. Setup Grafana to access kube-apiserver in order to fetch pod metrics
|
||
|
|
Plugin: https://grafana.com/plugins/grafana-kubernetes-app
|
||
|
|
Data Source: https://kubernetes.default.svc.cluster.local
|
||
|
|
Auth: With Credentials, With CA Cert, TLS Client Auth
|
||
|
|
|