Skip to content

Latest commit

 

History

History
154 lines (115 loc) · 2.89 KB

README.adoc

File metadata and controls

154 lines (115 loc) · 2.89 KB
Raw

Corteza k8s

This is an example of a Corteza deployment on a managed Kubernetes cluster.

Note

Since k8s cloud providers vary from one to another, some additional configuration per-provider is needed.
Important

This is a proof of concept kubernetes configuration for single-pod Corteza instance running with a default Postgresql installation.
What this configuration is lacking:

  • proper scaling support of Corteza server

  • no database operator or stateful set, db is running as a temporary store of value

  • no minio support

  • no tls (or letsencrypt)

  • no corredor (should be setup as a sidecar container)

The k8s recipes are tested on:
Corteza used in these examples

  • server (with webapps) 2022.9.5

  • postgresql 14.5

Prerequisites

A working node(s) and a control plane, configured networking plugin. Kubectl tools need to be setup locally in order to setup the cluster.

K8s node SSH

Dump kubeconfig and set as default
# merge conf
KUBECONFIG=~/.kube/config:~/my-test-cluster.kubeconfig k config view --flatten > ~/.kube/config

# alias kubectl
alias k=kubectl

Namespace

First create a separate namespace for Corteza
k create namespace corteza

Nodes

Update the inotify (inodes) limits on each of the nodes.

Important

The sysctl settings need to be higher, otherwise kubernetes starts failing.
Note

You can use the ssh kubectl plugin for access.
Ssh to the node
# fetch first node (but do not forget to do it for all)
NODE=$(k get nodes -o jsonpath='{.items[0].metadata.name}')

# ssh to node
k ssh node ${NODE}
Update sysctl (on node)
sysctl fs.inotify.max_user_instances=8192
sysctl fs.inotify.max_user_watches=524288

echo 'fs.inotify.max_user_instances=8192' > /etc/sysctl.conf
echo 'fs.inotify.max_user_watches=524288' > /etc/sysctl.conf
Create volume folders (on node)
# db volume
mkdir /mnt/data

# corteza volume
mkdir /mnt/corteza

Corteza service and deployments

Create a new Corteza deployment with services and configmaps
k apply -f server.yml -n corteza
Create a database deployment
k apply -f db.yaml -n corteza

Debug

DNS

Use the dnsutils for DNS debugging
# create pod
k apply -f dnsutils.yml -n corteza

# lookup service
k exec -it dnsutils -- nslookup postgres

Cleanup

Remove Corteza service
k delete -f server.yml -n corteza
Remove db service
k delete -f db.yml -n corteza
Remove the whole namespace if needed
k delete ns corteza