Upgrade and uninstall

This page upgrades a Hibernator release and removes one. The commands use the release name hibernator in the namespace hibernator, and the values file my-values.yaml from install.md.

• Upgrade
• Uninstall

Upgrade

1. Read what changed

Unpack the chart version that you want to install:

helm pull oci://registry.gitlab.com/cirriton/hibernator/charts/hibernator --version X.Y.Z --untar

Read hibernator/CHANGELOG.md. It lists the changes of each version. A line that starts with Upgrade: tells you what to do before or after the upgrade.

2. Upgrade the release

helm upgrade hibernator oci://registry.gitlab.com/cirriton/hibernator/charts/hibernator \
  -n hibernator -f my-values.yaml --version X.Y.Z

Without --version, Helm installs the newest version.

Give the same values file on every upgrade, and give each --set-file option again, such as the license or the brand logos. A value that you do not give goes back to its default. Without its license, Hibernator wakes everything it scaled down and stops scaling.

If you give the license file with --set-file, the command is:

helm upgrade hibernator oci://registry.gitlab.com/cirriton/hibernator/charts/hibernator \
  -n hibernator -f my-values.yaml --version X.Y.Z \
  --set-file license=HL-XXXXXXXX.license

Hibernator reads its configuration from a ConfigMap and reloads it when it changes. Most changes take effect without a pod restart.

From 0.12.0 on, Hibernator scales only where its license names the place. Before you upgrade an older release to 0.12.0 or later, set license: license.md says what to check.

Uninstall

Stand Hibernator down before you uninstall it. If you do not, the workloads that are hibernated at that moment stay scaled to zero: when the controller is gone, nothing wakes them.

1. Stand Hibernator down. As an admin, click Stand down in Manual Control, or send POST /api/v1/stand-down. Stop Hibernator in an emergency gives every way, step by step.

2. Wait until the stand-down shows that the handback is finished (how to see it). Hibernator has then woken everything it scaled down, resumed the CronJobs it suspended, started the databases it stopped and removed its satellites.

3. Uninstall the chart:

   helm uninstall hibernator -n hibernator

The stand-down stays in the hibernator-state ConfigMap, which the chart does not own. A reinstall into the same namespace therefore comes back stood down and names the admin who gave the order. It scales nothing until an admin ends the stand-down. To start from nothing, delete the ConfigMap after the uninstall:

kubectl delete configmap hibernator-state -n hibernator

config.global.enabled: false is a stand-down too, from the configuration. It hands back in the same way, and only a change of the value ends it.