Multi main n8n with zero downtime when you upgrade, is it possible?

[canob]

Hi Everyone,

We want to move from a single instance of n8n, to an infrastructure with multi main and queue mode.

My question is that, if we reach that point in a successful manner (we are reviewing our options here, Docker Swarm, K8S), is there a way to do upgrades on that infrastructure, without downtime?

Because, reviewing the multi main docs, I saw this:

• All main and worker processes are running the same version of n8n.

That is really annoying, thinking in that one of the motivations to change to a new infrastructure is that one, to get no downtime on upgrades, because commonly you have a new version of n8n every week…

My questions here are:

• Anyone with experience in this scenario of multi main and queue mode, how you manage the upgrades?
• Anyone is using this scenario on Docker Swarm? Any piece of advice or a recommended docker compose that I can use as the base for our deploy?

Thanks in advance for your help.

[Orionpax]

@canob You cannot upgrade n8n’s multi main cluster without stopping everything first. The reason is that all parts must run the same version at the same time, so you can’t update one piece while others are still running the old version.