This message was deleted.
# generals
sparse-intern-71089
07/04/2020, 11:10 AMThis message was deleted.
d
dazzling-sundown-39670
07/04/2020, 11:10 AM Copy code
export const cluster = new digitalocean.KubernetesCluster('cluster', {
region: digitalocean.Regions.LON1,
version: 'latest',
nodePool: {
name: 'default',
size: digitalocean.DropletSlugs.DropletS1VCPU2GB,
nodeCount: 2,
},
});dazzling-sundown-39670
07/04/2020, 11:11 AMI also tried destroying the stack and up-ing it again, same results when I run
pulumi previewdazzling-sundown-39670
07/04/2020, 11:11 AMAny suggestions?
b
broad-dog-22463
07/04/2020, 4:11 PM@dazzling-sundown-39670 I’d suggest setting the version explicitly rather than latest as that could result in upgrading at any point
g
green-school-95910
07/04/2020, 6:29 PMI think, might be wrong, that their API return the exact version that the cluster is running in including the patch number.
So when Pulumi refreshes the state prior to calculating the preview it see different strings (1.18/latest vs 1.18.x).
For the API setting 1.18 means the latest 1.18.x, which is already set, but for Pulumi those are different strings
If DigitalOcean applies patch updates automatically than maybe the provider should consider not updating x.y.z with x.y by checking the prefix. If they don't then the provider is correct it is trying to update to the latest patch every time.
Using latest will always result in a update request as there might be new k8s versions available.
d
dazzling-sundown-39670
07/07/2020, 8:23 AMYeah the solution was to do
version: '1.18.3-do.0',1.18dazzling-sundown-39670
07/07/2020, 8:23 AMCheers