https://pulumi.com logo
Join Slack
Powered by
I'd like to +1 to this issue, it's now blocking a ...
# general
e
I'd like to +1 to this issue, it's now blocking a lot of our infrastructure creation with EKS: https://github.com/pulumi/pulumi-kubernetes/issues/456
c
@early-musician-41645 I thought we fixed this, let me see what’s up
@early-musician-41645 This seems like it should be fixed with: https://github.com/pulumi/pulumi-kubernetes/pull/483
What version are you running?
d
Yeah I think we are running 0.21
c
@damp-book-35965 @early-musician-41645 I’ll try to reproduce the issue and the get something working.
g
@early-musician-41645 @damp-book-35965 I updated the issue. It looks like the issue was due to helm configuration, and not on our end. Let me know if you’re still having problems!
d
Got it..thanks. That solves the issue at hand but we need the CRD's for the helm chart to operate properly..
Thanks anyway @gorgeous-egg-16927
g
@damp-book-35965 IIUC, the operator will create the CRDs. the problem is that helm was also trying to create them
i.e. pulumi was failing because the prometheus-operator was creating the CRDs “out of band”
d
Yup I understood that..I think we will have to turn the option of here and use another chart to create the CRDs with dependsOn in pulumi
Thanks for looking into it
👍 1
c
I don’t think I understand… it should all “just work” if you turn the option off, no?
d
Nope..this will work but then CRD's need to be present in the cluster beforehand somehow..
c
The operator creates them, no?
d
If I set createCustomResource to false then no
c
I could be misunderstanding, but isn’t that instructing the Helm Chart to not create them?
The operator still should, right?
oh, perhaps you are right.
YOu don’t need a chart right. Just use 
k8s.yaml.ConfigFile
pointed at these URLs:
Copy code
<https://raw.githubusercontent.com/coreos/prometheus-operator/master/example/prometheus-operator-crd/alertmanager.crd.yaml>
<https://raw.githubusercontent.com/coreos/prometheus-operator/master/example/prometheus-operator-crd/prometheus.crd.yaml>
<https://raw.githubusercontent.com/coreos/prometheus-operator/master/example/prometheus-operator-crd/prometheusrule.crd.yaml>
<https://raw.githubusercontent.com/coreos/prometheus-operator/master/example/prometheus-operator-crd/servicemonitor.crd.yaml>
👍 1
d
So as part of the helm chart, it does the crd installs also which is breaking
c
I’m not sure what you mean. But, @gorgeous-egg-16927 tells me that I was right… 
createCustomResource
only stops the Chart from creating the CRDs. The operator will still do it?
Is that right, @gorgeous-egg-16927
g
Yes, just confirmed. You can check the prometheus-operator pod logs to confirm
d
Not sure how the operator creates the CRDs ?
Then why does the option even exist..it's redundant
g
the helm chart creates the operator deployment, which creates a pod, which creates the 4 CRDs
c
but, this is kind of what operators do, so it’s not super surprising to me.
g
note also that deleting the helm chart leaves the CRDs around. it seems pretty backwards to me
d
Yes
It's not a pulumi problem for sure
c
Helm does not really have a robust notion of lifecycles. That’s not surprising either.
g
this is why we only recommend using helm as a migration path, not as the end goal
d
Alright
s
@early-musician-41645 is always causing problems. (hi there 😄 )
😂 1
Open in Slack
PreviousNext
Powered by