Error getting VM curent VM status - Proxmox error

[abrondijk]

Relevant telegraf.conf:

# Provides metrics from Proxmox nodes (Proxmox Virtual Environment > 6.2).
[[inputs.proxmox]]
  ## API connection configuration. The API token was introduced in Proxmox v6.2. Required permissions for user and token: PVEAuditor role on /.
  base_url = "https://localhost:8006/api2/json"
  api_token = "<api_token>"
  ## Node name, defaults to OS hostname
  #node_name = ""

  ## Optional TLS Config
  # tls_ca = "/etc/telegraf/ca.pem"
  # tls_cert = "/etc/telegraf/cert.pem"
  # tls_key = "/etc/telegraf/key.pem"

  ## Use TLS but skip chain & host verification
  insecure_skip_verify = true
  # HTTP response timeout (default: 5s)
  response_timeout = "5s"

System info:

PVE 7.0-11 based on debian 11
Telegraf version 1.19.2, installed according to this guide.

Steps to reproduce:

1. Start the telegraf service

Expected behavior:

I expect telegraf to behave normally and to not spam errors in syslog.

Actual behavior:

The following error on every poll:

Aug 11 12:51:05 Raptor telegraf[625244]: 2021-08-11T10:51:05Z E! [inputs.proxmox] Error getting VM curent VM status: json: cannot unmarshal number into Go struct field VMStat.data.vmid of type string
Aug 11 12:51:05 Raptor telegraf[625244]: 2021-08-11T10:51:05Z E! [inputs.proxmox] Error getting VM stats: %vjson: cannot unmarshal number into Go struct field VMStat.data.vmid of type string

Besides the fact that current is spelled with only 1 R, I expect the data from proxmox to be gathered and sent to my output plugins.

Additional info:

I can confirm that the api token I use is correct, because if I change it, I get a completely different error.

[atetevoortwis]

I've managed to fix this by changing the type of the VM ID from string to int: https://github.com/atetevoortwis/telegraf/tree/bugfix/9611
With that change I can read stats from a Proxmox 7 host.
I don't know whether this int/string change is a Proxmox 6 vs 7 thing though...

[abrondijk]

That seems like a simple enough fix

I've managed to fix this by changing the type of the VM ID from string to int: https://github.com/atetevoortwis/telegraf/tree/bugfix/9611

@atetevoortwis Where are the proxmox scrips located? I can't seem to find the files to edit in my installation.

[atetevoortwis]

@abrondijk , see the only commit in my branch: atetevoortwis@9dd992f

[rare-magma]

@atetevoortwis I've tested your changes and it solves the problem for me as well. I'd suggest perhaps opening a pull request so this can get fixed?

[zeus86]

i can btw. confirm this error with the following versions:
pve-manager/7.0-11/63d82f4e (running kernel: 5.11.22-4-pve)
telegraf 1.20.0-1
another node with latest proxmox 6 and the same telegraf version with the same config runs fine.

[zeus86]

the issue btw. still remains, because the contribution license is not accepted in the PR...

[UniverseXXX]

I have 2 nodes and both not working.
Update: manual build as per above has fixed the issue.

[abrondijk]

You can just build telegraf yourself from the branch mentioned by atetevoortwis, that works fine for me on the latest proxmox 7.