pvesm: command not found

[shinn5112]

Hi all,

I have been trying to get set up over the weekend and have kept running into the following error when trying to set up a new VM:
Error: creating custom disk: environment: line 1: pvesm: command not found

The only thing I have seen that seems similar is caused by an issue with using ssh versus using the built-in console (see tteck/Proxmox#276). Not sure if my issue is related to a recent change to the provider, proxmox, or just a beginner error :)

Here is my provider file. Note that I am passing the credentials for the root@pam user for the node.

terraform {
    required_version = ">=1.5"
    required_providers {
        proxmox = {
            source = "bpg/proxmox"
            version = "0.68.1"
        }
    }
}

provider "proxmox" {
    endpoint  = var.proxmox_uri
    username  = var.proxmox_user
    password  = var.proxmox_password
    insecure  = var.proxmox_insecure
    tmp_dir   = "/var/tmp"

    ssh {
        agent = true
    }
}

Here is my resource file

data "local_file" "ssh_public_key" {
  filename = var.ssh_pubkey_path
}

resource "proxmox_virtual_environment_download_file" "debian_12_bookworm_qcow2_img" {
  content_type = "iso"
  datastore_id = "local"
  file_name    = "debian-12-genericcloud-amd64.qcow2.img"
  node_name    = "pve"
  url          = "https://cloud.debian.org/images/cloud/bookworm/latest/debian-12-genericcloud-amd64.qcow2"
}

resource "proxmox_virtual_environment_vm" "debian_test" {
  name        = "Podman"
  description = "Managed by Terraform"
  node_name   = "pve"
  tags        = ["terraform", "podman", "debian"]
  vm_id       = 200

  initialization {
    ip_config {
      ipv4 {
        address = var.podman_config.ipv4
        gateway = var.podman_config.gateway
      }
    }

    user_account {
      username = var.admin_account.username
      password = var.admin_account.password
      keys     = [trimspace(data.local_file.ssh_public_key.content)]
    }
  }

  disk {
    datastore_id = "local-lvm"
    file_id      = proxmox_virtual_environment_download_file.debian_12_bookworm_qcow2_img.id
    interface    = "virtio0"
    iothread     = true
    size         = var.podman_config.disk_size
  }

  cdrom {
    interface = "ide2"
  }

  network_device {
    bridge = "vmbr0"
  }

  cpu {
    cores = var.podman_config.cpu_cores
    type  = "x86-64-v2-AES"
  }

  memory {
    dedicated = var.podman_config.memory
  }

  operating_system {
    type = "l26"
  }
 
}

Does anything here look like I have done something wrong, or is this a potential bug?

Thanks in advance for any guidance you can provide :)

[bpg]

Hey @shinn5112 👋🏼

Looks like the shell the provider gets on the PVE node isn’t a root shell. Since you’re using an SSH agent, the shell might belong to a different user depending on your agent config, keys, or user setup. That user might not have access to /usr/sbin/pvesm on the PVE node.

Can you try SSHing into the PVE node directly from your terminal as root and running pvesm to see if it works?

Also, what’s the default shell for root on your PVE node? The provider only works with bash, so that could be part of the issue.

[shinn5112]

Hey @bpg ,

Thanks for the help! Turns out I wasn't running with the root@pam user. I forgot to save the changes to my tfvars file, so I was running with a different pam account. I guess I had been looking at the screen too long and made a simple mistake :/

Again, thanks for the help, and happy holidays! :)

[bpg]

Happy holidays!