Question: Purge task results older than X days

[nodesocket]

Question

Is there a way to automatically purge/delete task results older than X number of days? If this is exposed anywhere could use Semaphore to dogfood and create a scheduled task to run the purge.

Related to

No response

[DevAlphaKilo]

Dogfood all day baby!

This is available via the API and a programmatic approach. I currently do something similar by collecting the available tasks (last 200) and delete them via Ansible. My playbook can be found below, and with minor modifications you can achieve your goal. The variables are either populated in the vars section or from the vault. Then you can schedule this for w/e interval you desire. Additionally you could just go drop the rows from the backend database.

All joking aside, this would be a nice feature to have built-in to the software to help maintain the size of the database itself. Old jobs typically serve no purpose in my environment.

---

- name: Purge Ansible Semaphore Project Tasks (Last 200)
  hosts: (your-controller-host-here | all)

  vars:
    ansible_semaphore_host: (controller-hostname | controller-ip)
    ansible_semaphore_port: (controller-port)
    ansible_semaphore_project: '{{ project_id_input }}'

  tasks:

    - name: Login to Ansible Semaphore
      ansible.builtin.uri:
        url: "http://{{ ansible_semaphore_host }}:{{ ansible_semaphore_port }}/api/auth/login"
        method: POST
        status_code: 204
        headers:
          accept: application/json
          Content-Type: application/json
        body_format: json
        body: '{"auth": "local_admin","password": "{{ ansible_semaphore_admin_password }}"}'
      register: login_response

    - name: Get Last Tasks (Last 200)
      ansible.builtin.uri:
        url: "http://{{ ansible_semaphore_host }}:{{ ansible_semaphore_port }}/api/project/{{ ansible_semaphore_project }}/tasks/last"
        method: GET
        headers:
          accept: application/json
          Content-Type: application/json
          Cookie: "{{ login_response.cookies_string }}"
      register: tasks_last

    - name: "Display debug output for items"
      ansible.builtin.debug:
        msg: "Item: {{ item.status }}"
      when: not (item.status == "running" or item.status == "waiting")
      with_items:
        - "{{ tasks_last.json | list }}"

    - name: Remove Collected Tasks
      ansible.builtin.uri:
        url: "http://{{ ansible_semaphore_host }}:{{ ansible_semaphore_port }}/api/project/{{ item.project_id }}/tasks/{{ item.id }}"
        method: DELETE
        status_code: 204
        headers:
          accept: application/json
          Content-Type: application/json
          Cookie: "{{ login_response.cookies_string }}"
        timeout: 300
      when: not (item.status == "running" or item.status == "waiting")
      with_items:
        - "{{ tasks_last.json | list }}"
      loop_control:
        pause: '{{ delay_input | default(0) }}'

    - name: Logout of Ansible Semaphore
      ansible.builtin.uri:
        url: "http://{{ ansible_semaphore_host }}:{{ ansible_semaphore_port }}/api/auth/logout"
        method: POST
        status_code: 204
        headers:
          accept: application/json
          Content-Type: application/json
          Cookie: "{{ login_response.cookies_string }}"

[nodesocket]

Thanks for sharing the playbook, but agree this should be built-into Semaphore UI as a nightly cleanup job. That Ansible playbook is a big much to maintain for my liking to implement.

[nodesocket]

@DevAlphaKilo I found another potential solution though not clean at all. I installed boltbrowser on the physical host that is running the Semaphore UI docker container. Stopped the Semaphore UI docker container. Then from the physical host that is running the Semaphore UI container ran boltbrowser /var/lib/docker/volumes/semaphoreui-data/_data/database.boltdb.

From the boltbrowser cli tool, I was able to manually delete all entries inside the tasks bucks and all buckets with task output. Not clean like I said but it worked. Maybe helps others.

[DevAlphaKilo]

@DevAlphaKilo I found another potential solution though not clean at all. I installed boltbrowser on the physical host that is running the Semaphore UI docker container. Stopped the Semaphore UI docker container. Then from the physical host that is running the Semaphore UI container ran boltbrowser /var/lib/docker/volumes/semaphoreui-data/_data/database.boltdb.

From the boltbrowser cli tool, I was able to manually delete all entries inside the tasks bucks and all buckets with task output. Not clean like I said but it worked. Maybe helps others.

Hope this helps others, I prefer the API personally but have dropped rows in the database too for upgrades to get past issues, so great find! I completely agree with your initial suggestion, this would be a great built-in feature for the product to help reduce the footprint of the service.

[nodesocket]

@DevAlphaKilo I found another potential solution though not clean at all. I installed boltbrowser on the physical host that is running the Semaphore UI docker container. Stopped the Semaphore UI docker container. Then from the physical host that is running the Semaphore UI container ran boltbrowser /var/lib/docker/volumes/semaphoreui-data/_data/database.boltdb.
From the boltbrowser cli tool, I was able to manually delete all entries inside the tasks bucks and all buckets with task output. Not clean like I said but it worked. Maybe helps others.

Hope this helps others, I prefer the API personally but have dropped rows in the database too for upgrades to get past issues, so great find! I completely agree with your initial suggestion, this would be a great built-in feature for the product to help reduce the footprint of the service.

One caveat, BoltDB does not release disk space when doing deletes, so if you want the disk space back you have to manually run compact via the bbolt cli. Which is somewhat annoying.