Error installing EKS Provider
What happened?
I tried to run pulumi up using this action and I get an error preventing it from working.
error: could not load plugin for eks provider 'urn:pulumi:Cinder-Dev::cinder_app::pulumi:providers:eks::default': Could not automatically download and install resource plugin 'pulumi-resource-eks', install the plugin using `pulumi plugin install resource eks`.
Underlying error: 403 HTTP error fetching plugin from https://api.github.com/repos/pulumi/pulumi-eks/releases/latest
This worked before and stopped working this morning.
Expected Behavior
It to work.
Steps to reproduce
- Use EKS provider
- Deploy via Github Actions
Output of pulumi about
pulumi about
CLI
Version 3.53.1
Go Version go1.19.5
Go Compiler gc
Plugins
NAME VERSION
aws 5.26.0
aws 5.16.2
aws-native 0.46.0
awsx 1.0.1
cloudflare 4.15.0
command 0.7.0
docker 3.6.1
eks 1.0.1
kubernetes 3.23.1
nodejs unknown
random 4.8.2
tls 4.6.1
Host
OS ubuntu
Version 22.04
Arch x86_64
<removed for privacy>
Dependencies:
NAME VERSION
@pulumi/cloudflare 4.15.0
@pulumi/command 0.7.0
@pulumi/tls 4.6.1
ava 5.1.1
@pulumi/aws-native 0.46.0
@types/node 17.0.45
@pulumi/eks 1.0.1
@pulumi/pulumi 3.51.0
@pulumi/aws 5.26.0
@pulumi/awsx 1.0.1
@pulumi/kubernetes 3.23.1
@pulumi/random 4.8.2
@tsconfig/recommended 1.0.1
subnet-cidr-calculator 1.0.12
typescript 4.9.4
@aws-sdk/client-ec2 3.245.0
Pulumi locates its logs in /tmp by default
Additional context
No response
Contributing
No response
Hi Matt! Thanks for opening this issue. I have a few follow up questions for you:
- Was this a one-off error, or did it continue to fail after rerunning the Workflow?
- Were you able to reproduce this issue on the CLI? i.e. is this an error specific to GitHub Actions?
- Is it possible the error could be related to an expired or rate-limited GitHub token?
Underlying error: 403 HTTP error fetching plugin from https://api.github.com/repos/pulumi/pulumi-eks/releases/latest
The error message indicates this plugin is installed from GitHub's API. If your GitHub token is expired or rate limited, you'll be unable to download the plugin. This may be the source of the error.
This error is common enough for our users to see that I'd be surprised if it were specific to the GitHub Action. For my money, I'd bet the error would be caused by (3).
Hi Matt! Thanks for opening this issue. I have a few follow up questions for you:
- Was this a one-off error, or did it continue to fail after rerunning the Workflow?
It is very intermittent.
- Were you able to reproduce this issue on the CLI? i.e. is this an error specific to GitHub Actions?
Nope, it only happens under Github Actions.
- Is it possible the error could be related to an expired or rate-limited GitHub token?
Underlying error: 403 HTTP error fetching plugin from https://api.github.com/repos/pulumi/pulumi-eks/releases/latest
The error message indicates this plugin is installed from GitHub's API. If your GitHub token is expired or rate limited, you'll be unable to download the plugin. This may be the source of the error.
This error is common enough for our users to see that I'd be surprised if it were specific to the GitHub Action. For my money, I'd bet the error would be caused by (3).
I don't know how I would verify this. We have other pulumi based Workflows that don't use EKS that have never had a similar problem.
I am still struggling with this and it is causing issues with our deployments.
Do you have any specific guidance on how to fix this?
@RobbieMcKinstry The error message has changed but we are still running into this, in fact today none of our GH Actions have been able to execute.
error: could not load plugin for eks provider 'urn:pulumi:Cinder-Dev::cinder_app::pulumi:providers:eks::default': Could not automatically download and install resource plugin 'pulumi-resource-eks', install the plugin using `pulumi plugin install resource eks`.
Underlying error: 401 HTTP error fetching plugin from https://api.github.com/repos/pulumi/pulumi-eks/releases/latest
Wouldn't a possible solution be to host your plugins on a CDN?
Hi I am facing the same situation under my pipeline in the last month. Same behavior and error message