kopsrox

usage

• image
  • create
  • destroy
  • info
• cluster
  • create
  • update
  • info
  • kubectl
  • kubeconfig
  • destroy
• etcd
  • snapshot
  • list
  • restore
  • prune

image

create <a name=image-create>

• downloads the image file defined in koprox.ini as up_image_url under the [proxmox] section
• installs qemu-guest-agent into the image via virt-customise
• imports the disk into the proxstore proxmox storage using sudo qm
• creates cloudinit drive with user and networking setup
• converts the vm into a template
• patched cloudimage is backed up

destroy <a name=image-destroy>

• :warning: deletes the existing image template
• delete the .img file manually ( eg mantic-minimal-cloudimg-amd64.img if you want a fresh download/repatch of the upstream image

info <a name=image-info>

• prints info about image/template vm eg storage, id, creation time and source cloud image file

cluster

create <a name=cluster-create>

• creates and updates a cluster - use this to setup a fresh cluster
• clones the image to kopsrox-m1 master server and configures it via cloudinit
• exports kubeconfig and node token
• if an existing working master is found just runs the same steps as kopsrox cluster update

update <a name=cluster-update>

• checks the state of the cluster vs what is configured in kopsrox.ini
• adds or deletes workers/masters per kopsrox.ini

info <a name=cluster-info>

• shows a list of ids, hostnames and ips the host they are running on
• shows kubectl get nodes

kubectl

• provides a quick and basic way to run some kubectl commands for example:

./kopsrox.py cluster kubectl get events -A

kubeconfig

• export the cluster kubeconfig to a $cname.kubeconfig file - ie the name of the cluster set in kopsrox.ini
• file is patched to have correct master IP vs 127.0.0.1

destroy <a name=cluster-destroy>

• :warning: destroys the cluster ( NO WARNING! )
• deletes workers then masters in safe order

etcd

snapshot

The first time a snapshot is taken the cluster token is written to clustername.etcd.token

This is not overwriten when further snapshots are taken - even on a new cluster with the same name

./kopsrox.py etcd snapshot

Takes a backup of etcd

./kopsrox.py etcd list

Should show the new backup

list <a name=s3-list>

./kopsrox.py etcd list

• lists snapshots taken in s3 storage based on cluster name

restore

restores a cluster from a snapshot

usage:

./kopsrox.py etcd restore $imagename

• :white_check_mark: check you’re using the correct kopsrox.etcd.snapshot.token file for the snapshot!

• downsizes the cluster to 1 node then resizes back to the scale set in kopsrox.ini

prune

• deletes old snapshots per the retention policy set on the bucket