From 98ccc2742dd291f4e07d55994ff3ed5340b77e7c Mon Sep 17 00:00:00 2001 From: Marco Nenciarini Date: Mon, 21 Mar 2022 19:12:04 +0100 Subject: [PATCH] Improve stability of traefik load balancer IP By default, if the helm chart installation or upgrade fails for any reason, the whole helm chart is deleted and recreated from scratch, with a new load balancer bound to a new IP. By setting the traefik helm chart failure policy to `stop`, we prevent the load balancer service from being recreated in case of chart upgrade failure, at the cost of requiring manual user intervention to restore automatic upgrades. --- templates/traefik_config.yaml.tpl | 1 + 1 file changed, 1 insertion(+) diff --git a/templates/traefik_config.yaml.tpl b/templates/traefik_config.yaml.tpl index 03fd6e7..6cd89eb 100644 --- a/templates/traefik_config.yaml.tpl +++ b/templates/traefik_config.yaml.tpl @@ -4,6 +4,7 @@ metadata: name: traefik namespace: kube-system spec: + failurePolicy: stop valuesContent: |- service: enabled: true