https://pulumi.com logo
Join Slack
Powered by
It appears that changing an Helm release `values` ...
# kubernetes
d
It appears that changing an Helm release 
values
(Go automation API) doesn’t trigger an update during Up() operations. I can see the new value in the stack state file but not when I run 
helm get values <release>
. I have tried setting Replace=true / ForceUpdate=true / ResetValues=true but no luck. Am I doing something wrong?
b
cc @sparse-park-68967
Could you open an issue with a repro please @dry-autumn-28966
s
As @billowy-army-68599 mentioned, could you open an issue? I don't know off the top of my head. Would be helpful to look at the details. We have tests covering updates which pass FWIW so this might be specific to the chart in question?
d
@billowy-army-68599 / @sparse-park-68967 https://github.com/pulumi/pulumi-kubernetes/issues/1754
s
Hey @dry-autumn-28966 - looking at the issue, are you actually using nginx chart? Are you sure there is a postgresql.custom block in the values for that chart? I believe Helm just ignores values not in the chart.
FWIW I tried messing around this with another chart locally and set values from config. Things worked just fine - as long as the values existed in the chart.
d
@sparse-park-68967 Oops, I was originally using postgres then switched to Nginx. The issue occurs regardless of the chart though. It stops updating the values as soon as I call the Up with Target() against the Stack URN. I am going to send you a quick 3 mins video in private. This time I use a “real” value that changes the port of the postgresql service
@sparse-park-68967 So you’re not seeing the same behavior as in the video I sent?
s
I haven't had a chance to look through it, apologies. I should be able to get to it in a couple of hours.
🙏 1
So looks like the first update worked fine. The second update didn't take?
So this doesn't seem to be specific to updates (normal updates work fine). It seems to be exacertbated around refresh followed by update. I think I can investigate in the morning. The series of operations you are doing are a little weird though. Why the target up on the stack (which skips the update of the helm release resource)?
Some more context around what you are trying to achieve would be helpful
d
@sparse-park-68967 I have a 
k8s
stack that is responsible for generating a KubeConfig file and create a few k8s resources (Namespace / Secrets). The kubeconfig that is generated has a token that is only going to be valid for a couple of hours. My 
app
stack requires the KubeConfig file from 
k8s
via StackReference to deploy an Helm chart and various k8s resources… Now this is where things get tricky: • I want to refresh the 
app
stack before running up • Pulumi doesn’t refresh providers during refresh, so it may run with an expired KubeConfig and fail. The only solution I found is to run the following actions: • Run up against 
k8s
to refresh KubeConfig • Run targeted up against 
app
stack’s 
Stack
and 
K8s provider
resources. • Run 
app
refresh • Run 
app
up It works fine but then I hit the Helm issue https://github.com/pulumi/pulumi-kubernetes/issues/1754
Were you able to look into it?
s
Apologies - have been busy with another release. We released a few improvements to helm release recently. I am curious if things improve with the v3.8.1? I intend to look into the issue you filed this week. Thanks for your patience
d
Same issue with both latest pulumi and Kubernetes plugin 😕
@sparse-park-68967 Any update?
5 Views
Open in Slack
PreviousNext
Powered by