[Docs] Workspaces RAM and CPU requirements

[l0rd]

Is your enhancement related to a problem? Please describe.

Users don't know how much RAM and CPU a workspace will need. They don't know if a workspace will fail to start or not in their cluster because of the memory or cpu limits.

We pre-calculate the memory limits for the defaults stacks (e.g. for angular it's 2686Mi). But we haven't documented how to calculate these limit for a custom stack.

Describe the solution you'd like

Write a document (under Using developer workspaces -> Workspaces memory and CPU requirements) that should answer the following question:

• Given a devfile, how much RAM the corresponding workspace will need (the memory limit)?
• Given a existing k8s cluster resource quotas and limit ranges policies how can I figure out if my workspace will be scheduled or not?

Write a document (under Administration Guide -> Cluster memory and CPU requirements) that answer the following question:

• Given a sample workspace, and a number of users that should be able to use the workspace concurrently, what should be the cluster capacity and the suggested suggested resource quotas and limit ranges?

[slemeur]

Linking to downstream issue: https://issues.jboss.org/browse/CRW-228

[nickboldt]

@l0rd @rkratky is this something we might get done for Che 7.9 / CRW 2.1? I know at least one customer who's looking for guidance ...

[l0rd]

@nickboldt @themr0c @rkratky I have created the doc, I will seek for some reviews by TLs (and of course yours as well) and then I will transform it in asciidoc and publish it. Anyway it should be definately ready for 2.1.

[themr0c]

fixed by eclipse-che/che-docs#1107