Hi,
I’ve started testing out your brilliant rancher!
I got a small issue. Not a serious one, but something worth discussing
I have a Crate.io stack (including LB++) running on vagrant/ranceros. If I disable my stack and reload the vagrant cluster, all services are disabled except from crate which is started.
Is this supposed to work like this? The nodes are ultimatly failing since no dns records are updated.
I’m not sure I completely understand your set up. Can you provide more details on how you have your setup? Rancher server/agents, how many agents, maybe share your docker-compose of your stack (hide anything that’s private)?
There is no way to “disable a stack”. This is assuming you are talking about the stack inside Rancher which is under the tab Applications/Stacks.
Where is the vagrant cluster and how are you “reloading it”
Hi, and thanks for answering.
I was maybe a little bit unclear.
I’m using the os-vagrant having 4 rancheros hosts
I have a stack in rancher with different services including load balancers
I stack menu I click stop services, and all containers are stopped.
then I do vagrant halt
vagrant up
When all hosts are up again, my load balancers services are not running (also container are stopped)
However, my other services are also stopped(inactive) but the service containers are running.
I would think that containers in a stack should not be booted up when the service is inactive.
PS, service has restart: always so i guess thats why service is starting…
It’s because your service has a restart:always, which would cause the containers to start. Since the restart policy is “always”, it means that if the host goes down and comes back up, the containers will automatically start again (default Docker behavior when the restart policy is set to alway). But since Rancher knows that you had placed the service into “inactive”, it keeps that state in the UI.
I can put a bug to say it’s weird user experience, but I am not exactly sure what the exact solution should be.