Cannot create a Helm chart - timeout during pulumi up update preview phase

[JinLisek]

What happened?

I try to create a Kubernetes cluster on Proxmox VM (created via Pulumi).

When I run pulumi up I need to wait a while until kubernetes resources show up in preview, I mean these:

When they finally show up, after about 10 seconds the pulumi preview fails with exception:

Exception: invoke of kubernetes:helm:template failed: invocation of kubernetes:helm:template returned an error: failed to generate YAML for specified Helm chart: could not get server version from Kubernetes: Get "https://192.168.1.101:6443/version?timeout=32s": dial tcp 192.168.1.101:6443: i/o timeout

192.168.1.101 is ip address of my master node that is not yet created (it should get created during pulumi up).

I made it so kubernetes provider depends on command that downloads kube config from master node. Namespace depends on the provider, and helm chart depends on the namespace... So I would expect the helm chart to request version from master node during pulumi up, when it's time to create the chart... Not during the preview .

Example

import ipaddress

import pulumi
import pulumi_command as command
import pulumi_kubernetes as kubernetes
import pulumi_proxmoxve as proxmox

...

master_node_ip = ipaddress.IPv4Interface(address="192.168.1.101/24")

provider = proxmox.Provider(
    resource_name="proxmoxve-provider",
    endpoint="https://192.168.1.69:8006",
    username="root@pam",
    password=password,
    insecure=True,
    ssh=proxmox.ProviderSshArgs(agent=True, private_key=pve_authorized_private_key),
)

# creates proxmox.vm.VirtualMachine, details not important
master_node = create_virtual_machine(
    provider=provider,
    name="lol-k8s-master-0001",
    password=password,
    authorized_keys=[vm_authorized_public_key],
    ip_interface=master_node_ip,
)

# install and setup master node using kubeadm init
setup_kubernetes_master = command.local.Command(
    resource_name="ansible-setup-for-lol-k8s-master-0001",
    create=master_node.ipv4_addresses.apply(
        lambda ips: f"ansible-playbook ansible/setup_kubernetes_master.yaml --inventory {get_first_remote_ip_address(ips)}, --vault-password-file ./password.txt"
    ),
)

# download kube config locally
copy_kube_config_command = command.local.Command(
    resource_name="copy-kube-config",
    create=f"scp -r lol@{master_node_ip.ip}:/home/lol/.kube/ ~/",
    delete="rm ~/.kube/ -r",
    opts=pulumi.ResourceOptions(
        depends_on=setup_kubernetes_master,
        custom_timeouts=pulumi.CustomTimeouts(create="10s"),
    ),
)

# note that kubernetes provider depends on copy_kube_config_command 
kubernetes_provider = kubernetes.Provider(
    resource_name="k8s",
    kubeconfig="~/.kube/config",
    opts=pulumi.ResourceOptions(depends_on=copy_kube_config_command),
)

tigera_operator_namespace = kubernetes.core.v1.Namespace(
    resource_name="tigera-operator-namespace",
    metadata=kubernetes.meta.v1.ObjectMetaArgs(name="tigera-operator"),
    opts=pulumi.ResourceOptions(provider=kubernetes_provider),
)


kubernetes.helm.v3.Chart(
    release_name="calico",
    config=kubernetes.helm.v3.LocalChartOpts(
        path="./helm/calico/tigera-operator", namespace="tigera-operator"
    ),
    opts=pulumi.ResourceOptions(depends_on=tigera_operator_namespace),
)

Output of pulumi about

CLI
Version 3.113.3
Go Version go1.22.2
Go Compiler gc

Plugins
KIND NAME VERSION
resource command 0.10.0
resource kubernetes 4.11.0
resource proxmoxve 6.4.1
language python unknown

Host
OS ubuntu
Version 22.04
Arch x86_64

This project is written in python: executable='/home/lol/repos/infra/.venv/bin/python3' version='3.12.1'

Current Stack: lolol/infra/dev

Found no resources associated with dev

Found no pending operations associated with dev

Backend
Name pulumi.com
URL https://app.pulumi.com/lolol
User lolol
Organizations lolol
Token type personal

Dependencies:
NAME VERSION
ansible 9.5.1
mypy 1.10.0
pip 23.3.1
pulumi_command 0.10.0
pulumi_kubernetes 4.11.0
pulumi_proxmoxve 6.4.1

Pulumi locates its logs in /tmp by default

Additional context

When I comment out the kubernetes stuff and run pulumi up, Pulumi successfully setups my master node - all pods (except coredns) are running, node is NotReady because no CNI installed yet.
After that I can uncomment kubernetes resources, run pulumi up again and it will successfully deploy Calico (after which my master node becomes Ready).

I cannot switch to using Helm Release, because it also doesn't work well - when I try to pulumi destroy resources, the command gets stuck on deleting the Release.

Contributing

Vote on this issue by adding a 👍 reaction.
To contribute a fix for this issue, leave a comment (and link to your pull request, if you've opened one already).