Background
Today, a lot of nodes were rebooted within the same 1 hour window. No two nodes overlapped, and the upgrade went smoothly. However, if such an upgrade would fail we would have had little time to react before the whole network would have been down.
While we have #3462 to ensure certain upgrades do not restart the CVMs, it also makes sense to ensure we spread out these automatic upgrades as much as possible.
Therefore, we should provide a script/tool that helps operators pick a slot in the week for automatic upgrades and configure the auto-upgrades to only happen at that time. Ideally 1/5 of operators would upgrade on Mondays, 1/5 on Tuesdays etc.
User Story
As a node operator I don't want my node to restart at the same time as other nodes.
Acceptance Criteria
We've provided a tool to help node operators coordinate their auto-upgrades to achieve maximum spread of when these happen.
Resources & Additional Notes
No response
Background
Today, a lot of nodes were rebooted within the same 1 hour window. No two nodes overlapped, and the upgrade went smoothly. However, if such an upgrade would fail we would have had little time to react before the whole network would have been down.
While we have #3462 to ensure certain upgrades do not restart the CVMs, it also makes sense to ensure we spread out these automatic upgrades as much as possible.
Therefore, we should provide a script/tool that helps operators pick a slot in the week for automatic upgrades and configure the auto-upgrades to only happen at that time. Ideally 1/5 of operators would upgrade on Mondays, 1/5 on Tuesdays etc.
User Story
As a node operator I don't want my node to restart at the same time as other nodes.
Acceptance Criteria
We've provided a tool to help node operators coordinate their auto-upgrades to achieve maximum spread of when these happen.
Resources & Additional Notes
No response