DNS Latency probe: IPv6 DNS servers cause duplicate metric collision

[svengerber]

Description

After upgrading from v1.24.1 to v1.25.0 (DNS Latency metric added in #334), probing fails with a duplicate metric error when the FortiGate has IPv6 DNS servers configured.

Error

An error has occurred while serving metrics:
collected metric "fortigate_network_dns_latency_seconds" { label:{name:"ip" value:"32.1.22.32"} label:{name:"service" value:"dns_server"} gauge:{value:0.01}} was collected before with the same name and label values

Root Cause

The FortiGate obtains two IPv6 DNS servers via DHCPv6:

2001:1620:2777:1::10
2001:1620:2777:2::20

The exporter appears to truncate these to IPv4, resulting in both being mapped to the same ip label value(32.1.22.32 in hex is 20:01:16:20 → which is 2001:1620). This causes a duplicate label collision.

Environment

Exporter version: v1.25.0 (worked fine on v1.24.1, which didn't have the DNS Latency probe)
FortiGate: FortiOS 7.4.11
WAN interface: DHCP + DHCPv6

Workaround

Either exclude the probe in fortigate-key.yaml:

probes:
  exclude:
    - System/DNSLatency

Or disable IPv6 DNS on the FortiGate (disable "Override internal DNS" on the WAN interface).

[bastischubert]

hey @svengerber
do you mind sending me the full (redacted) output for the api endpoint that causes this? this helps creating a better testcase 🧪

[svengerber]

Thank you for looking into this @bastischubert

Here's what the FortiGate API actually returns (tested against FortiOS 7.4.11):

curl -sk -H "Authorization: Bearer XXXX" "https://X.X.X.X/api/v2/monitor/network/dns/latency?vdom=root" -k

{
  "http_method": "GET",
  "results": [
    {
      "service": "dns_server",
      "latency": 10,
      "last_update": 17180,
      "ip": "32.1.22.32",
      "ip6": "2001:1620:2777:2::20"
    },
    {
      "service": "dns_server",
      "latency": 10,
      "last_update": 46910,
      "ip": "32.1.22.32",
      "ip6": "2001:1620:2777:1::10"
    },
    {
      "service": "dns_server",
      "latency": 10,
      "last_update": 17010,
      "ip": "149.112.112.112"
    },
    {
      "service": "dns_server",
      "latency": 10,
      "last_update": 17180,
      "ip": "9.9.9.9"
    }
  ],
  "vdom": "root",
  "path": "network",
  "name": "dns",
  "action": "latency",
  "status": "success",
  "serial": "FWF60EXX00000000",
  "version": "v7.4.11",
  "build": 2878
}

[bastischubert]

TBH the API is returning data in a way that looks weird to me, looks like the fortigate itself does the "magic" of converting 20:01:16:20 to 32.1.22.32 - this clearly looks like an error to me

Since there is NO proper documentation on that API (in the forti developer network) i'd like to ask you if you can open a ticket with fortinet that explains the results here or fixes that ambigous output.