Instead it should replace just the node that was modified...
07/26/2019, 4:05 AM
That certainly isn’t expected?
Are you using a separate NodePool resource? I know that that is generally much more reliable way to keep these independent with GKE.
07/26/2019, 12:18 PM
Would be happy to Hangouts with you and show you what I'm doing and behavior I'm seeing
And no I'm not, I'll try that as well
I'm not sure this works either
also replace is completely broken...
it leaves the old cluster completely intact after a replacement is created...
07/26/2019, 7:47 PM
I imagine that your replacement failed? There deletion of the old cluster will happen at the end of the replacement (to avoid downtime), but if there is a failure it will be processed on the next update.
If you would like to find some time for a screenshare to show what you are seeing - I'd be happy to find some time this afternoon - reach out in DM and we can coordinate.
07/26/2019, 8:09 PM
im still testing and tweaking
see if this one goes smoother
Can't for the life of me get it to work using a seperate nodepool..If I combine it all together under cluster and use NodePools array it works fine unless I modify one of the NodePools then it attempts to do a full cluster replacement.
its apparently possible to have a completely empty cluster in gcp
which makes me think I should be able to setup an empty cluster and then designate two nodes on it after the fact
What's a little odd though is you can't create a cluster with 0 nodes as the creation process prevents it. But you can delete all of your nodes after the fact and have it. I am wondering if thats intended by Google or a bug.