[Internal]: Elasticsearch - Documenting full restart-upgrades of clusters #3879
Description
Description
Currently, the documentation only covers the rolling upgrade path as a viable option to upgrade a cluster. However, there's a mention here, here & here that aside from rolling upgrades, a full restart-upgrade is an option too. A full restart-upgrade should be documented as a viable option should a user acknowledge the risks and downtime associated with such an upgrade, as it would mean to take all nodes in a cluster offline, upgrade them & start them again.
Resources
It seems this might have been documented in the past (see description) and is a viable option to upgrade a cluster.
- This might be best documented as a sub-category here:
- A disclaimer should be presented that a full restart-upgrade means down time as well as additional risks
- This sort of upgrade should only be considered in specific (development or test) use-cases, and discouraged to be performed in production environments
Which documentation set does this change impact?
Elastic On-Prem and Cloud (all)
Feature differences
On Elastic Cloud, a rolling upgrade is the only way to upgrade a cluster. This could also be revisited in a sense that if a user would opt for a full cluster upgrade-restart, it could potentially be given as an option, if a user would acknowledge risks + downtime.
What release is this request related to?
N/A
Serverless release
N/A
Collaboration model
Other (please describe below)
Point of contact.
Main contact: @emrcbrn
However this could potentially have some cross-engineering requirements to implement a full upgrade-restart method in Cloud, aside from documenting this option for on-prem deployments, so leaving it as other.
Action
added by (@georgewallace)
- Review the guidance on full restart upgrade from issue
- Validate with engineering if this is a supported way we want to add to the upgrade section
- Determine plan to implement guidance