Hi all, I’m working on migrating our K8s manifests...
# general
s
Hi all, I’m working on migrating our K8s manifests from being managed by Helm to Pulumi, everything is working great, except I’m having trouble importing the service objects into pulumi without requiring a recreate strategy on the first up. These are just normal cluster ip services, and the only difference that I can tell from the state file is that pulumi wants to remove the actual
clusterIP
state. Obviously, this isn’t usually tracked in an IaC template as it’s dynamically set, and so I’m confused why the recreate strategy keeps getting chosen. I’d really like to make this switch over zero downtime, so looking for a solution here, or wondering if anyone else has run into this before. Thanks!
b
an you share an example diff?
you might need to import the existing service so it keeps the ip
s
let me see if i still have one sticking around. this is after importing the existing service. the only diff from up was the
clusterIP
stuff, so i have to assume that’s what is driving the recreate
@billowy-army-68599
like annotations and labels shouldn’t cause recreate, so i can only assume it’s these ips
wondering what happens if i manually delete these from the state file
b
is the urn changing? i think its cut off that image
s
yeah the provider urn? it is
but it’s changing on all my resources that are being updated not replaced
okay yeah, if i manually remove the ips from the state file, it switches to an update as i’d expect
just seems like these maybe shouldn’t be tracked on import
so that’s a solution for me — this does seem like a bug to me as otherwise this would cause downtime on importing, happy to make a ticket if you agree
b
agree it sounds like a bug, would you mind filing an issue in pulumi-kubernetes?
1