With multiple Doco CD instances in same repository, can I prevent all of them triggers the scheduler?

[iguana31]

I configured two instances of Doco CD in one repository for self-updating purposes, as suggested in #827 (comment). It works very well so far.

I also have a Renovate instance in the same repository which is triggered by Doco CD's scheduler. It worked great while I had only one Doco CD instance. Now, with a second instance, each of them triggers at the same time, leading to two conflicting Renovate containers running in parallel.

I tried to configure a separate target for one of the Doco CD instances and I also tried to set the label cd.doco.job.skip_running: true in the Renovate Compose file. Neither approach fixed the problem.

Is it somehow possible to prevent one Doco CD instance from triggering the scheduler? Otherwise I will use an external trigger for Renovate. Although I like the idea to use the Doco CD scheduler very much.

[kimdre]

Do you use the same deployment config for both doco-cd instances? I would recommend to write a additional deployment config just for the doco-cd instance that updated the main/app doco-cd instance.

[iguana31]

Yes, I have two separate deployment configs (.doco-cd.main.yaml and .doco-cd.updater.yaml). Only the main target contains the deployment for renovate. For the updater, polling is configured with target: updater. But both instances are recognizing the labels in the renovate compose file and scheduling the job.

I'm doing experiments with Compose profiles, but I didn't find a satisfying and stable solution yet.

[kimdre]

Can you please show me the configs and compose files?

[iguana31]

Sure. Thanks for your support, I really appreciate!

.doco-cd.main.yaml

name: doco-cd_updater
reference: feature/doco-cd_updater_target 
working_dir: ./doco-cd_updater
force_recreate: true
external_secrets:
  # external secrets ...

---
name: renovate
reference: feature/doco-cd_updater_target
working_dir: ./renovate
external_secrets:
  # external secrets ...

.doco-cd.updater.yaml

name: doco-cd
reference: feature/doco-cd_updater_target 
working_dir: ./doco-cd
external_secrets:
  # external secrets ...

doco-cd/compose.yaml

services:
  app:
    container_name: doco-cd
    image: ghcr.io/kimdre/doco-cd:0.95.0
    restart: unless-stopped
    expose:
      - 8034  # Webhook endpoint
    # For all available environment variables and explanations, see https://doco.cd/latest/App-Settings/
    environment:
      TZ: Europe/Berlin
      HTTP_PORT: 8034  # use a non-privileged port
      SECRET_PROVIDER: webhook
      SECRET_PROVIDER_WEBHOOK_STORES_FILE: /secret-store.yml
      GIT_ACCESS_TOKEN: ${DOCOCD_TOKEN_FORGEJO}
      WEBHOOK_SECRET: ${DOCOCD_WEBHOOOK_SECRET}
      LOG_LEVEL: DEBUG
    volumes:
      - ./secret-store.yml:/secret-store.yml:ro
      - /var/run/docker.sock:/var/run/docker.sock
      - data:/data
    networks:
      - bw-api
      - doco-cd
    cap_drop:
      - ALL
    healthcheck:
      test: [ "CMD", "/doco-cd", "healthcheck" ]
      start_period: 15s
      interval: 30s
      timeout: 5s
      retries: 3
    depends_on:
      bitwarden-api:
        condition: service_healthy

  bitwarden-api: # sidecar that provides the Bitwarden Vault Management API
    image: ghcr.io/kimdre/bitwarden-rest-api-server:2026.5.0
    environment:
      # Refer to the image documentation for the configuration options: https://github.com/kimdre/bitwarden-rest-api-server#getting-started
      - BW_HOST=${BW_HOST}
      - BW_CLIENTID=${BW_CLIENTID:?error}
      - BW_CLIENTSECRET=${BW_CLIENTSECRET:?error}
      - BW_PASSWORD=${BW_PASSWORD:?error}
      - VAULT_SYNC_INTERVAL=3600
    expose: 
      - 8087
    networks:
      - bw-api
    restart: unless-stopped
    init: true
    read_only: true
    cap_drop:
      - ALL
    tmpfs:
      - /data:uid=65532,gid=65532,mode=755

volumes:
  data:
    name: doco-cd_data

networks:
  bw-api:
    name: doco-cd_bw-api
  doco-cd:
    name: doco-cd

doco-cd_updater/compose.yaml

services:
  app:
    container_name: doco-cd_updater
    image: ghcr.io/kimdre/doco-cd:0.95.0
    restart: no
    # For all available environment variables and explanations, see https://doco.cd/latest/App-Settings/
    environment:
      TZ: Europe/Berlin
      HTTP_PORT: 8000  # use a non-privileged port
      SECRET_PROVIDER: webhook
      SECRET_PROVIDER_WEBHOOK_STORES_FILE: /secret-store.yml
      GIT_ACCESS_TOKEN: ${DOCOCD_TOKEN_FORGEJO}
      POLL_CONFIG_FILE: /poll-config.yaml
      LOG_LEVEL: DEBUG
    volumes:
      - ./poll-config.yaml:/poll-config.yaml:ro
      - ./secret-store.yml:/secret-store.yml:ro
      - /var/run/docker.sock:/var/run/docker.sock
      - data:/data
    networks:
      - bw-api
    cap_drop:
      - ALL
    healthcheck:
      test: [ "CMD", "/doco-cd", "healthcheck" ]
      start_period: 15s
      interval: 30s
      timeout: 5s
      retries: 3

volumes:
  data:
    name: doco-cd_updater_data

networks:
  bw-api:
    name: doco-cd_bw-api
    external: true

doco-cd_updater/poll-config.yaml

- url: # URL of my repository
  reference: feature/doco-cd_updater_target
  run_once: true
  target: updater

renovate/compose.yaml

services:
  renovate:
    container_name: renovate
    image: docker.io/renovate/renovate:43.269.2
    restart: no
    environment:
      RENOVATE_PLATFORM: forgejo
      RENOVATE_ENDPOINT: # API endpoint of my repository
      RENOVATE_TOKEN: ${RENOVATE_TOKEN_FORGEJO:-}
      RENOVATE_AUTODISCOVER: false
      RENOVATE_REPOSITORIES: # List containing the name of my repository
      RENOVATE_GITHUB_COM_TOKEN: ${RENOVATE_GITHUB_TOKEN:-}
    labels:
      cd.doco.job.enabled: true
      cd.doco.job.schedule: "0 * * * *"
      cd.doco.job.execution_mode: one_off
      cd.doco.job.skip_running: true

[kimdre]

I think I know the issue.
The scheduler is not target-aware and discovers all services/containers with cd.doco.job.enabled, so both instances pick up the Renovate job. skip_running only prevents duplicate runs within a single Doco CD process, not across multiple instances.

I need to implement something to prevent this.

[kimdre]

Can you please test with 0.97.0-rc.1? Set the image tag for both instances to 0.97.0-rc.1 and then add the SCHEDULER_ENABLED: false environment variable to the updater instance to disable the job scheduler there.

[iguana31]

Thanks, it looks very good.

The logs:

doco-cd                  | {"time":"2026-07-18T20:53:40.498229448+02:00","level":"info","msg":"starting scheduler","component":"scheduler"}
doco-cd                  | {"time":"2026-07-18T20:53:40.501662483+02:00","level":"info","msg":"job scheduled","component":"scheduler","job":"renovate","mode":"container","next_run":"2026-07-18T21:00:00+02:00","schedule":"0 * * * *"}
doco-cd                  | {"time":"2026-07-18T20:53:40.592422451+02:00","level":"warning","msg":"new application version available","current":"v0.97.0-rc.1","latest":"v0.96.0"}
doco-cd                  | {"time":"2026-07-18T21:00:05.00032359+02:00","level":"info","msg":"scheduled run completed","component":"scheduler","execution_mode":"one_off","job":"renovate","job_id":"019f7699-8391-7d27-a0cf-211d942c8f2e","mode":"container","next_run":"2026-07-18T22:00:00+02:00","scheduled_at":"2026-07-18T21:00:00+02:00","stack":"renovate"}

doco-cd_updater  | {"time":"2026-07-18T20:53:20.681791052+02:00","level":"info","msg":"scheduler disabled by configuration"}
doco-cd_updater  | {"time":"2026-07-18T20:53:21.198009767+02:00","level":"warning","msg":"new application version available","current":"v0.97.0-rc.1","latest":"v0.96.0"}

And now, only one "doco-job" container is being created at the scheduled time.

[iguana31]

One note to your well-written new Self-Updating documentation:
In the example, you use compose_file: compose.main.yaml. Shouldn't it be compose_files: ["compose.main.yaml"] instead?

[kimdre]

Yes, you are right. :D I will fix it.