update_to_24.04.md

Update Optscale to new release

(Update to Ubuntu 24.04 LTS Noble Numbat, k8s, BuildKit, nerdctl)

⚠️ Before pulling new release

• make sure you backed up data folder/volume /optscale
• make sure you backed up user overlays ~/optscale/optscale-deploy/overlay/

In this example instruction data directory is placed in /optscale, sources placed in ~/optscale

1. navigate to optscale-deploy repo:

$ cd ~/optscale/optscale-deploy/

2. activate .venv

~/optscale/optscale-deploy$ source .venv/bin/activate

3. delete deployment

~/optscale/optscale-deploy$ ./runkube.py -d  -- <release_name> local

if you're forgotten release name, you can use this command:

$ helm list | grep optscale

The output should be like:

$ helm list | grep optscale
NAME         	REVISION	UPDATED                 	STATUS  	CHART               	APP VERSION	NAMESPACE
os-1      	1       	Thu Dec 26 12:32:35 2024	DEPLOYED	optscale-0.1.0      	           	default

according to the output, my release is: os-1 and the command will be:

~/optscale/optscale-deploy$ ./runkube.py -d  -- os-1 local

12:29:10.285: Deleting optscale cluster os-1 on k8s 172.24.1.14
release "os-1" deleted

4. deactivate and remove virtual environment

~/optscale/optscale-deploy$ deactivate
~/optscale/optscale-deploy$ rm -r .venv/

5. reset k8s

⚠️ All Kubernetes applications, including the k8s agent, will be deleted and must be reinstalled if necessary.

$ sudo kubeadm reset

$ rm $HOME/.kube/config

6. remove k8s, docker

$ sudo apt remove --purge kube*
$ sudo apt remove --purge docker-*

7. update base system

$ sudo apt update && sudo apt dist-upgrade -y && sudo apt clean && sudo apt autoclean && sudo apt autoremove -y

8. reboot server

$ sudo reboot

9. do release upgrade

get version:

$ lsb_release -a

update manager core and do release upgrade:

$ sudo apt install update-manager-core

20.04 -> 22.04

$ sudo do-release-upgrade

answer 'y' and follow instructions

22.04 -> 24.04

$ lsb_release -a

...
Description:	Ubuntu 22.04.5 LTS
Release:	22.04

$ sudo do-release-upgrade

answer 'y' and follow instructions

10. install requirements

$ sudo apt update; sudo apt install python3-pip sshpass git python3-virtualenv python3

11. pull new version

~/optscale$ git fetch && git pull

12. create new venv

~/optscale$ cd optscale-deploy/

~/optscale/optscale-deploy$ virtualenv -p python3 .venv

~/optscale/optscale-deploy$ source .venv/bin/activate

~/optscale/optscale-deploy$ pip install -r requirements.txt

13. run ansible playbook

~/optscale/optscale-deploy$  ansible-playbook -e "ansible_ssh_user=<user>" -k -K -i "<ip>," ansible/k8s-master.yaml

when user - your actual user ip - our cluster ip address

14. re-login or source ~/.profile

$  source ~/.profile

15. (optional) build new components

cd .. && ./build.sh --use-nerdct

16. (optional) recover your user overlays from backup

⚠️ The overlay needs access to your cluster data saved in your data folder, for example, database passwords, service account, etc. typically default overlay is stored in user_template.yml

17. start updated cluster with saved overlay

./runkube.py --no-pull --with-elk  -o overlay/user_template.yml -- <release_name> <version>

when release_name - your cluster name version - your version, for example I want to run cluster os-1 with locally built version:

./runkube.py --no-pull --with-elk  -o overlay/user_template.yml -- os-1 local