Zone/Hosts Outputs not generating, but Inputs events are happening

[BismuthumOS]

Summary

Hi! So I'm trying to run herald in compose, with host Docker Daemon input, to generate probably .zone file out of it (So I can give it to dnsmasq probably). I want to dynamicly get DNS records of my running containers in that Daemon.

Steps to reproduce

/output/zone and /output/hosts have chmod 666.
I do have logs in debug mode when I restart some containers on my Daemon.
But my output files not being touched at all. I do not have warnings or errors about it :(

What is the expected correct behavior?

Zone and Hosts files must have DNS records

Relevant logs and/or screenshots

herald.yml

outputs:
  zone:
    type: file
    format: zone
    path: "/output/zone"
    domain: local
    soa:
      primary_ns: ns1.local
      admin_email: admin@local
      serial: auto
    ns_records:
      - ns1.local
      - ns2.local
  
  hosts:
    type: file
    format: hosts
    path: "/output/hosts"
    domains: ["example.com"]
    # DNS Resolution Control Options:
    flatten_cnames: true          # Enable CNAME to A record flattening (default: true)
    dns_server: "1.1.1.1"        # Use external DNS server (Cloudflare)
    resolve_external: true       # Force external DNS resolution
    ip_override: "192.168.1.100" # Override all resolved IPs with this address`

domains:
  domain_01:
    name: local
    profiles:
      inputs:
        - docker_input
      outputs:
        - zone
        - hosts
    record:
      type: A
      target: "server.example.com"  # Default target for new records
      ttl: 300
      update_existing: true
      allow_multiple: false

Herald Logs: Input Docker

2025-11-27 10:14:44     INFO [docker/connection-manager/tcp://172.17.0.1:2375] Started shared event streaming
2025-11-27.10:14:44 [STARTING] ** [scheduling] Starting cron
2025-11-27 10:15:48     INFO [input/docker/docker_input] Container stopped: herald-test (8a5df7d48f2c), removing DNS records for this container only
2025-11-27 10:15:48     INFO [input/docker/docker_input] Container started: herald-test (8a5df7d48f2c), processing DNS records for this container only
2025-11-27 10:15:48     INFO [input/docker/docker_input] Container stopped: crazy_kalam (c2c09f512f5d), removing DNS records for this container only
2025-11-27 10:15:48     INFO [input/docker/docker_input] Container started: crazy_kalam (c2c09f512f5d), processing DNS records for this container only

Herald Logs: Init config

2025-11-27 10:14:44     INFO [config] Using config file: /config/herald.yml
2025-11-27 10:14:44     INFO [output] Registered output profile 'hosts' (hosts)
2025-11-27 10:14:44     INFO [output] Registered output profile 'hosts_file' (hosts)

Environment

• Image version / tag: nfrastack/herald:latest, Herald version: 2.3.1
• Host OS: Debian 12

Any logs | docker-compose.yml

services:
  herald:
    image: docker.io/nfrastack/herald
    environment:
      - DEBUG_MODE=TRUE
      - HERALD_SETUP_TYPE=manual
    volumes:
      - ./herald/herald.yml:/config/herald.yml:ro
      - ./herald/logs:/logs
      - ./herald/output:/output
    restart: unless-stopped

[tiredofit]

Ineresting. Can you show me your input configuration? You match pretty closely how I am using it for split-dns purposes.

Thoughts in the dark right now:

• I don't think we'd be in a permissions issue for writing to output - Could potentially try HERALD_USER=root to cancel out any container permissions issues.

• LOG_LEVEL=DEBUG or TRACE is going to reveal a whole bunch of info but may tell us whats going on here.

• I see two host configs there, and no zone, I assume it was just testing and omitted from the issue config sharing. That's fine, I already validated and all looks good for me. The only thing that differs ddrastically from my own config would be this: path: /zonefiles/%domain%.zone I also don't have a domain entry in my zone config.

[BismuthumOS]

Thank you for your support! My inputs:

inputs:
  docker_input:
    type: docker
    api_url: tcp://172.17.0.1:2375
    # api_url: unix:///var/run/docker.sock
    process_existing: true
    expose_containers: true
    swarm_mode: false
    record_remove_on_stop: false
    tls:
      verify: false

I do want to use tcp:// socket, but I checked mounting docker socket anyway, thinking that tcp socket IS the problem. I had strange permissions issues with docker.socket, so I gave it up and continue rewriting config :)

Just to be sure, I added HERALD_USER=root, LOG_LEVEL=DEBUG and removed domain field.

2025-11-28 05:51:12  VERBOSE [docker/shared] Container event: 'stop' - name: 'nginx_app1' - id: 'bccf1dd15ae2'
2025-11-28 05:51:12  VERBOSE [docker/shared] Container event: 'die' - name: 'nginx_app1' - id: 'bccf1dd15ae2'
2025-11-28 05:51:12  VERBOSE [docker/shared] Container event: 'start' - name: 'nginx_app1' - id: 'bccf1dd15ae2'
2025-11-28 05:51:12    DEBUG [input/docker/docker_input] Processing container 'nginx_app1' because 'expose_containers=true' in config
2025-11-28 05:51:12     INFO [input/docker/docker_input] Container stopped: nginx_app1 (bccf1dd15ae2), removing DNS records for this container only
2025-11-28 05:51:12    DEBUG [input/docker/docker_input] Processing container 'nginx_app1' because 'expose_containers=true' in config
2025-11-28 05:51:12    DEBUG [input/docker/docker_input] Processing container 'nginx_app1' because 'expose_containers=true' in config
2025-11-28 05:51:12     INFO [input/docker/docker_input] Container started: nginx_app1 (bccf1dd15ae2), processing DNS records for this container only
2025-11-28 05:51:12    DEBUG [input/docker/docker_input] Removing DNS records for container: nginx_app1 (bccf1dd15ae2)
2025-11-28 05:51:12    DEBUG [input/docker/docker_input] No hostname/domain found for container 'nginx_app1', skipping
2025-11-28 05:51:12    DEBUG [input/docker/docker_input] No DNS entries found for container nginx_app1 cleanup
2025-11-28 05:51:12    DEBUG [input/docker/docker_input] Processing specific container: nginx_app1 (bccf1dd15ae2)
2025-11-28 05:51:12    DEBUG [input/docker/docker_input] Processing container 'nginx_app1' because 'expose_containers=true' in config
2025-11-28 05:51:12    DEBUG [input/docker/docker_input] Processing container 'nginx_app1' because 'expose_containers=true' in config
2025-11-28 05:51:12    DEBUG [input/docker/docker_input] No hostname/domain found for container 'nginx_app1', skipping

At first, I thought that I had no container_name in my compose config, but that wasn't true.

[tiredofit]

I see that for some reason it's processing your containers twice, that's not good. I think I sae something about it maybe on my own tests, but we'll have to dig in a bit deeper to see if this is actually the case.

So, are you using any labels when starting your container?

[BismuthumOS]

... no?
No labels :)

[tiredofit]

OK, You'll need some labels or a way to generate some output for your zones/hosts.
Also, your domains we'll want to setup better:

• change your `domain_01.name=example.com
• Add a label to your containers either:

nfrastack.herald.host: "test2.example.com

OR

traefik.http.routers.example_container.rule="Host(`test.example.com`)

Both syntax will work even if you aren't using Traefik, as this was meant to be a shim from another earlier tool I wrote years ago.

Herald can see your containers, and wants to act on it, but it doesn't know how to act on it so passing it a label with a hostname and the domain name (which we have setup in the updated config) will allow it to start processing and outputting records for you.