Optimize and reduce resources required
Summary
The current deployment of our Helm chart requires more resources than are available in the standard GKE cluster. By default, automatic cluster scaling is also disabled. This means if you just accept the standard cluster settings and try to deploy GitLab, it's not going to work.
While we should continue to reduce and tune the requests we make and our pod counts (for example 3x nginx controller), with an HA deployment and the amount of memory sidekiq and unicorn need I'm not sure we will get there. To solve this problem, we should provide an easy option to allow GitLab to be deployed.
I'd suggest some type of demo
flag, where it is clear we don't really recommend this for production use, but it is there for convenience.
Edited by Joshua Lambert