solo-io / gloo

The Cloud-Native API Gateway and AI Gateway
https://docs.solo.io/
Apache License 2.0
4.09k stars 446 forks source link

Add tip about upgrading federated edge #6078

Open Rachael-Graham opened 2 years ago

Rachael-Graham commented 2 years ago

Version

master

Describe the requested changes

Question came up about whether there are specific guidelines for upgrading gloo-fed and gloo-edge, especially in multi-cluster environments where it will take some time to upgrade all edges. Example provided: Let’s imagine I have two clusters with federated gloo edge. I upgrade gloo edge in cluster1. This new release might imply updating CRDs or creating new ones. How that affects the other gloo edge in cluster 2? Can federation work with 2 different versions? Upgrading the federation, how does that impact the federated gloo edges?

Answer: The general process for upgrading federated is pretty similar to gloo edge. First need make sure all CRDs have been updated by hand, and then install the new gloo version. In a multicluster case, we want to ensure that the "update CRDs" step is done to all relevant clusters before we proceed to the next one.

Link to any relevant existing docs

Considering this is not vastly different from the currently-documented upgrade steps, we'd need to either just add a tip to the existing upgrade page at https://docs.solo.io/gloo-edge/latest/operations/upgrading/upgrade_steps/ or we could potentially just add a section to the FAQs page at https://docs.solo.io/gloo-edge/latest/operations/upgrading/faq/

Browser Information

No response

Additional Context

No response

guydc commented 2 years ago

related: https://github.com/solo-io/gloo/issues/6063

guydc commented 2 years ago

@Rachael-Graham

  1. Should the Federation/Edge CRDs and Gloo components be deployed in a specific order?
  2. What about CRs? Should they be deployed before/after Gloo components are updated?
  3. Is there a clear statement regarding compatibility? Can Gloo Federation work with two different Gloo Edge versions at the same time? For example, consider a canary deployment, where new Gloo Edge instances are being created and traffic is gradually being shifted from older instances. Different edge versions will co-exist for a significant period of time.
  4. Are these steps sufficient to achieve Zero Downtime upgrades?
github-actions[bot] commented 5 months ago

This issue has been marked as stale because of no activity in the last 180 days. It will be closed in the next 180 days unless it is tagged "no stalebot" or other activity occurs.