Hi,
I wonder if ther’s a way to have with Rancher CLI the same behavior than the API when we upgrade a service in a stack : the possibility to start before stop.
In other words : is it possible to have zero downtime during a rancher CLI force-upgrade of a stack?
Thank you
I found something “almost” working : Adding this in the rancher-compose of the stack :
> upgrade_strategy:
> start_first: true
My problem is that when I create a new stack with this upgrade strategy and when I get back the docker-compose.yml and the rancher-compose.yml from API, this option has not been saved in it.
I’am using Rancher-CLI through Jenkins in my CI/CD workflow. And when I want to force an upgrade of my stack, my Jenkins Job starts with an API call to get the YAML files and then calls the rancher CLI upgrade. But if I open the rancher-compose.yaml, I did not find the upgrade_startegy properties.
Is it a bug?
When you say through the API, are you performing a stack config export? Could be a bug here. Do you mind filing a bug so we can track this?
Yes, I get the stack config (a zip file with the docker-compose and the rancher-compose in it).
Do i have to create an issue on github?