From b2c0aa7ab93495e1313e24b9f08ef96f342c7682 Mon Sep 17 00:00:00 2001 From: Patrice LACHANCE Date: Sat, 16 May 2020 09:09:56 +0200 Subject: [PATCH] FAQ:Zero downtime images #397 taken from https://github.com/cloudposse/docs/pull/397/files --- .../faq/how-achieve-zero-downtime-image.md | 25 +++++++++++++++++++ 1 file changed, 25 insertions(+) create mode 100644 content/faq/how-achieve-zero-downtime-image.md diff --git a/content/faq/how-achieve-zero-downtime-image.md b/content/faq/how-achieve-zero-downtime-image.md new file mode 100644 index 000000000..e8b696cf7 --- /dev/null +++ b/content/faq/how-achieve-zero-downtime-image.md @@ -0,0 +1,25 @@ +--- +title: "How do we achieve image updates with zero downtime?" +description: "This is similar to scaling a cluster vertically." +tags: +- VM +- image +- kops +- updates +--- + +## Question + +How do we achieve zero-downtime deploys of VM image updates? + + +## Answer + +This is similar to scaling a cluster vertically. This can be done with zero downtime. + +Basically, it requires setting the `KOPS_BASE_IMAGE` env to the new value, rebuilding the `kops` manifest, then redeploying by running the `kops rolling-update` command. See [here.](https://github.com/cloudposse/geodesic/blob/master/rootfs/templates/kops/default.yaml#L171) + +Our instructions are [here.](https://docs.cloudposse.com/kubernetes-optimization/scale-cluster-vertically/) + +We don’t presently build the `kops` images. Using Packer would be the strategy we would take. +