You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
this will likely be long lasting ticket but "As an admin I want to uninstall the applications and framework from a cluster".
Just wanted to write it down here. Currently I found it easier to rebuild the cluster than hunt down all the stuff operators leave behind. And that's a problem, not all operators clean themseves up nicely. It might require creating an ansible playbook to go and delete all the resources that are left behind, and naturally file tickets to operators while doing so. Hopefully the list would shorten while operators enhance.
Now if the Makefile would be ansible, it might be easier to just mark stuff with state: absent/present syntax, which then would help keeping track what was created and what needs to be removed. But to begin with, perhpaps having some sort of nuke_it_all.yaml, which would include individual playbooks to remove a given feature. Also perhaps having a directory where to drop a playbook to clean up a given application user might have added. E.g. call the relevant helm uninstalls.
I don't know how you handle tasks, but I suppose this could be epic, and then each component might be separate task to clean up.
Also workflow doc would need section on this.
Consider for the use case an admin tries this in some real environment. And wants to remove it, and there are tons of other real company test/dev environments there, one definately can't just nuke the clusters this framework installed stuff. There needs to be a way to uninstall this so one isn't doomed after trial/error.
@ikke-t Thanks! We have this in our TODO list. You point out some of the challenges that we face with operators not cleaning up their resources. I will add this to our team meeting to discuss. It might be that we have an uninstall process for each pattern. At any rate thanks for opening the issue.
Hi,
this will likely be long lasting ticket but "As an admin I want to uninstall the applications and framework from a cluster".
Just wanted to write it down here. Currently I found it easier to rebuild the cluster than hunt down all the stuff operators leave behind. And that's a problem, not all operators clean themseves up nicely. It might require creating an ansible playbook to go and delete all the resources that are left behind, and naturally file tickets to operators while doing so. Hopefully the list would shorten while operators enhance.
Now if the Makefile would be ansible, it might be easier to just mark stuff with state: absent/present syntax, which then would help keeping track what was created and what needs to be removed. But to begin with, perhpaps having some sort of nuke_it_all.yaml, which would include individual playbooks to remove a given feature. Also perhaps having a directory where to drop a playbook to clean up a given application user might have added. E.g. call the relevant helm uninstalls.
I don't know how you handle tasks, but I suppose this could be epic, and then each component might be separate task to clean up.
Also workflow doc would need section on this.
Consider for the use case an admin tries this in some real environment. And wants to remove it, and there are tons of other real company test/dev environments there, one definately can't just nuke the clusters this framework installed stuff. There needs to be a way to uninstall this so one isn't doomed after trial/error.
Here is an example of k8s module in ansible creating/deleting resource using state. All resources in Fevermap OCP CI/CD were created with ansible, so it's not the best example for this as such, but it was 3.11 ocp time.
(I mistakenly created the ticket to docs site first, so i convert that just to docs ticket for the same issue)
The text was updated successfully, but these errors were encountered: