An Ansible role for aggregating logs from different nodes.
Go to file
Zuul b7430bdece Merge "Adjust regex for tripleo container image prepare log" 2023-09-15 23:31:31 +00:00
docs Bumping black to 22.3.0 2022-03-30 13:44:46 +02:00
infrared_plugin Add artcl_find_* as infrared arguments 2022-01-05 10:04:17 +00:00
plugins Fix ansible sanity and linters jobs 2022-11-15 13:09:26 +01:00
roles/collect_logs Merge "Adjust regex for tripleo container image prepare log" 2023-09-15 23:31:31 +00:00
test-playbooks Transform artcl into a collection 2021-01-29 11:34:14 +00:00
tests Fix ansible sanity and linters jobs 2022-11-15 13:09:26 +01:00
zuul.d Fix CI issues 2023-09-15 22:30:47 +02:00
.ansible-lint Fix CI issues 2023-09-15 22:30:47 +02:00
.envrc Convert flatten_nested_dict filter into a module 2020-04-09 11:32:14 +01:00
.gitignore Upgrade ansible-lint/molecule runs 2021-07-30 11:11:22 +01:00
.gitreview OpenDev Migration Patch 2019-04-19 19:44:00 +00:00
.pre-commit-config.yaml Fix CI issues 2023-09-15 22:30:47 +02:00
LICENSE Apply cookiecutter on collect-logs 2019-04-15 16:27:27 +05:30 Upgrade ansible-lint/molecule runs 2021-07-30 11:11:22 +01:00
README.rst Make rst files collection compatible 2021-02-03 17:37:14 +00:00
ansible.cfg Some linting, and correcting deps for sanity 2022-05-30 14:14:45 +02:00
bindep.txt Assure pyyaml is installed before running sova test 2020-02-20 17:14:58 +00:00
galaxy.yml Upgrade ansible-lint/molecule runs 2021-07-30 11:11:22 +01:00
requirements.txt Replace ansible requirement by ansible-core 2021-11-25 08:36:52 +00:00
requirements.yml Upgrade ansible-lint/molecule runs 2021-07-30 11:11:22 +01:00
setup.cfg Fix elinter isort 2023-04-14 13:00:58 +02:00 Disable auto discovery as a workaround 2022-03-25 18:25:53 +05:30
test-requirements.txt Use unittest.mock instead of mock 2020-11-27 13:31:27 +00:00
tox.ini Fix CI issues 2023-09-15 22:30:47 +02:00



Ansible role for aggregating logs from different nodes.

The only supported way to call this role is using its main entry point. Do not use tasks_from as this count as using private interfaces.


This role gathers logs and debug information from a target system and collates them in a designated directory, artcl_collect_dir, on the localhost.

Additionally, the role will convert templated bash scripts, created and used by TripleO-Quickstart during deployment, into rST files. These rST files are combined with static rST files and fed into Sphinx to create user friendly post-build-documentation specific to an original deployment.

Finally, the role optionally handles uploading these logs to a rsync server or to an OpenStack Swift object storage. Logs from Swift can be exposed with os-loganalyze.

Role Variables

File Collection

  • artcl_collect_list A list of files and directories to gather from the target. Directories are collected recursively and need to end with a '/' to get collected. Should be specified as a YaML list, e.g.:
    - /etc/nova/
    - /home/stack/*.log
    - /var/log/
  • artcl_collect_list_append A list of files and directories to be appended in the default list. This is useful for users that want to keep the original list and just add more relevant paths.

  • artcl_exclude_list A list of files and directories to exclude from collecting. This list is passed to rsync as an exclude filter and it takes precedence over the collection list. For details see the 'FILTER RULES' topic in the rsync man page.

  • artcl_exclude_list_append A list of files and directories to be appended in the default exclude list. This is useful for users that want to keep the original list and just add more relevant paths.

  • artcl_collect_dir A local directory where the logs should be gathered, without a trailing slash.

  • collect_log_types - A list of which type of logs will be collected, such as openstack logs, network logs, system logs, etc. Acceptable values are system, monitoring, network, openstack and container.

  • artcl_gzip: Archive files, disabled by default.

  • artcl_rsync_collect_list - if true, a rsync filter file is generated for rsync to collect files, if false, find is used to generate list of files to collect for rsync. find brings some benefits like searching for files in a certain depth (artcl_find_maxdepth) or up to certain size (artcl_find_max_size).

  • artcl_find_maxdepth - Number of levels of directories below the starting points, default is 4. Note: this variable is applied only when artcl_rsync_collect_list is set to false.

  • artcl_find_max_size - Max size of a file in MBs to be included in find search, default value is 256. Note: this variable is applied only when artcl_rsync_collect_list is set to false.

  • artcl_commands_extras - A nested dictionary of additional commands to be run during collection. First level contains the group type, as defined by collect_log_types list which determines which groups are collected and which ones are skipped.

    Defined keys will override implicit ones from defaults artcl_commands which is not expected to be changed by user.

    Second level keys are used to uniqly identify a command and determine the default output filename, unless is mentioned via capture_file property.

    cmd contains the shell command that would be run.

      cmd: df
      # will save output to /var/log/extras/dist-space.log
      cmd: mount -a
      capture_file: /mounts.txt  # <-- custom capture file location
      cmd: touch /foo.txt
      capture_disable: true # <-- disable implicit std redirection
      when: "1 > 2"  # <-- optional condition
  • artcl_gen_docs: false/true If true, the role will use build artifacts and Sphinx and produce user friendly documentation (default: false)
  • artcl_docs_source_dir a local directory that serves as the Sphinx source directory.
  • artcl_docs_build_dir A local directory that serves as the Sphinx build output directory.
  • artcl_create_docs_payload Dictionary of lists that direct what and how to construct documentation.
    • included_deployment_scripts List of templated bash scripts to be converted to rST files.
    • included_deployment_scripts List of static rST files that will be included in the output documentation.
    • table_of_contents List that defines the order in which rST files will be laid out in the output documentation.
  • artcl_verify_sphinx_build false/true If true, verify items defined in artcl_create_docs_payload.table_of_contents exist in sphinx generated index.html (default: false)
    - undercloud-install
    - undercloud-post-install
    - env-setup-virt
    - env-setup-virt
    - undercloud-install
    - undercloud-post-install
  • artcl_publish: true/false If true, the role will attempt to rsync logs to the target specified by artcl_rsync_url. Uses BUILD_URL, BUILD_TAG vars from the environment (set during a Jenkins job run) and requires the next to variables to be set.
  • artcl_txt_rename: false/true rename text based file to end in .txt.gz to make upstream log servers display them in the browser instead of offering them to download
  • artcl_publish_timeout: the maximum seconds the role can spend uploading the logs, the default is 1800 (30 minutes)
  • artcl_use_rsync: false/true use rsync to upload the logs
  • artcl_rsync_use_daemon: false/true use rsync daemon instead of ssh to connect
  • artcl_rsync_url rsync target for uploading the logs. The localhost needs to have passwordless authentication to the target or the PROVISIONER_KEY var specified in the environment.
  • artcl_use_swift: false/true use swift object storage to publish the logs
  • artcl_swift_auth_url the OpenStack auth URL for Swift
  • artcl_swift_username OpenStack username for Swift
  • artcl_swift_password password for the Swift user
  • artcl_swift_tenant_name OpenStack tenant (project) name for Swift
  • artcl_swift_container the name of the Swift container to use, default is logs
  • artcl_swift_delete_after The number of seconds after which Swift will remove the uploaded objects, the default is 2678400 seconds = 31 days.
  • artcl_artifact_url An HTTP URL at which the uploaded logs will be accessible after upload.
  • artcl_report_server_key - A path to a key for an access to the report server.
  • ara_enabled: true/false - If true, the role will generate ara reports.
  • ara_overcloud_db_path: Path to ara overcloud path (tripleo only).
  • ara_generate_html: true/false - Generate ara html.
  • ara_graphite_prefix: Ara prefix to be used in graphite.
  • ara_only_successful_tasks: true/false - Send to graphite only successfull tasks.
  • ara_tasks_map: Dictionary with ara tasks to be mapped on graphite.

Logs parsing

"Sova" module parses logs for known patterns and returns messages that were found. Patterns are tagged by issues types, like "infra", "code", etc. Patterns are located in file sova-patterns.yml in vars/ directory.

  • config - patterns loaded from file
  • files - files and patterns sections match
  • result - path to file to write a result of parsing
  • result_file_dir - directory to write a file with patterns in name

Example of usage of "sova" module:

- name: Run sova task
    config: "{{ pattern_config }}"
      console: "{{ ansible_user_dir }}/workspace/logs/quickstart_install.log"
      errors: "/var/log/errors.txt"
      "ironic-conductor": "/var/log/containers/ironic/ironic-conductor.log"
      syslog: "/var/log/journal.txt"
      logstash: "/var/log/extra/logstash.txt"
    result: "{{ ansible_user_dir }}/workspace/logs/failures_file"
    result_file_dir: "{{ ansible_user_dir }}/workspace/logs"

Example Role Playbook

- name: Gather logs
  hosts: all:!localhost
    - collect_logs
** Note:

The tasks that collect data from the nodes are executed with ignore_errors. For example:

Templated Bash to rST Conversion Notes

Templated bash scripts used during deployment are converted to rST files during the create-docs portion of the role's call. Shell scripts are fed into an awk script and output as restructured text. The awk script has several simple rules:

  1. Only lines between ### ---start_docs and ### ---stop_docs will be parsed.
  2. Lines containing # nodoc will be excluded.
  3. Lines containing ## :: indicate subsequent lines should be formatted as code blocks
  4. Other lines beginning with ## <anything else> will have the prepended ## removed. This is how and where general rST formatting is added.
  5. All other lines, including shell comments, will be indented by four spaces.

Enabling sosreport Collection

sosreport is a unified tool for collecting system logs and other debug information. To enable creation of sosreport(s) with this role, create a custom config (you can use centosci-logs.yml as a template) and ensure that artcl_collect_sosreport: true is set.

Sanitizing Log Strings

Logs can contain senstive data such as private links and access passwords. The 'collect' task provides an option to replace private strings with sanitized strings to protect private data.

The 'sanitize_log_strings' task makes use of the Ansible 'replace' module and is enabled by defining a sanitize_lines variable as shown in the example below:

  - dir_path: '/tmp/{{ inventory_hostname }}/etc/repos/'
    file_pattern: '*'
    orig_string: '^(.*)download(.*)$'
    sanitized_string: 'SANITIZED_STR_download'
  - dir_path: '/tmp/{{ inventory_hostname }}/home/zuul/'
    file_pattern: '*'
    orig_string: '^(.*)my_private_host\.com(.*)$'
    sanitized_string: 'SANITIZED_STR_host'

The task searches for files containing the sensitive strings (orig_string) within a file path, and then replaces the sensitive strings in those files with the sanitized_string.

Usage with InfraRed

Run the following steps to execute the role with infrared.

  1. Install infrared and add ansible-role-collect-logs plugin by providing the url to this repo:

    (infrared)$ ir plugin add --src-path infrared_plugin
  2. Verify that the plugin is imported by:

    (infrared)$ ir plugin list
  3. Run the plugin:

    (infrared)$ ir ansible-role-collect-logs


Apache 2.0

Author Information