Clean up the OIM¶

Use each domain's cleanup workflow to remove the services and artifacts that it owns. After the required domain cleanups succeed, remove the shared Omnia execution environment with the Main cleanup command.

Cleanup runs from the Omnia Infrastructure Manager (OIM). Keep the shared virtual environment installed until every required domain cleanup completes.

Danger

Cleanup is destructive. Depending on the selected domain and options, it can remove containers, repositories, images, credentials, cluster configuration, telemetry workloads, persistent data, backups, and shared-storage content. Back up all required data before continuing.

When to use OIM cleanup¶

Reset a failed or experimental deployment.
Remove selected Omnia domains before redeployment.
Return a lab or test OIM to a clean state.
Remove the shared Omnia environment after domain services are removed.

Prerequisites¶

Log in to the OIM as a user with the privileges required by every selected cleanup workflow.
Use the same OMNIA_PROJECT_NAME and resolved domain data paths used for deployment. In particular, preserve ORCHESTRATOR_DATA_PATH when Orchestrator used a custom root; when it is unset, Orchestrator uses <OMNIA_DATA_PATH>/orchestrator.
Confirm that the Omnia virtual environment is available.
Stop or drain workloads that use the services or storage being removed.
Back up project inputs, credentials, repository content, images, telemetry data, databases, log archives, configuration backups, and shared-storage data that must be retained.
Review the cleanup guide for each deployed domain before selecting optional data deletion or credential preservation behavior.

Cleanup scope by domain¶

Domain	Default cleanup scope
`build_stream`	Removes GitLab, BuildStreaM services, the watcher, PostgreSQL service, runtime artifacts, and BuildStreaM credentials. PostgreSQL data is preserved by default.
`telemetry`	Removes all enabled telemetry sources and sinks and deletes the stored telemetry credential file. Source-owned persistent volumes are removed; Kafka, VictoriaMetrics, and VictoriaLogs volumes are preserved by default.
`orchestrator`	Removes enabled OpenCHAMI, OpenLDAP, Slurm, Kubernetes, storage-mount, and generated Orchestrator resources. OpenCHAMI cleanup removes its packages and three persistent service volumes. The workflow prompts independently before deleting Slurm and Kubernetes shared data and removes credentials by default; use `-e cleanup_credentials=false` to preserve them.
`discovery`	Removes the current project's Discovery output contents and credentials while preserving the output directory and other staged inputs.
`image_build_manager`	Removes MinIO when locally managed, the image registry, build artifacts, domain runtime data, logs, and Image Build Manager credentials.
`repo_manager`	Removes the Pulp deployment, Pulp data, repository integration, logs, and Repository Manager credentials. Credentials and logs are removed by default.
`utils`	Removes cluster-log artifacts, unattended-installation temporary files and credentials, all OIM log-backup runs, and all Slurm configuration backup runs.

To preserve Image Build Manager services and remove only selected or all built artifacts, use Clean up built images instead of the full image_build_manager cleanup tag.

Clean up deployed domains¶

Run only the following sections for domains that have deployed or generated state. Keep the listed reverse dependency order so consumers are removed before the services they depend on. For example, remove Telemetry before its Kubernetes environment and remove Image Build Manager before Repository Manager.

Complete full-cleanup command sequence¶

Use the following block only when every listed domain was deployed and a complete data reset is intended. If a domain was not deployed, remove its command from the block before running it. The subshell exits immediately if a command fails, which prevents cleanup from continuing to a dependency or the shared environment.

Danger

This sequence deletes PostgreSQL data, all Telemetry source and sink volumes, and Slurm and Kubernetes shared data. It then removes the Omnia environment and all remaining data under OMNIA_DATA_PATH. Confirm that required data has been backed up outside every cleanup target.

Image Build Manager cleanup does not delete objects from an external PowerScale S3 provider. Review and remove external objects separately only when their deletion is intended.

Run on: OIM host

(
  set -e
  cd <OMNIA_SOURCE_PATH>/src/main

  ./omnia.sh --run build_stream --tags cleanup \
    -e postgres_backup=false
  ./omnia.sh --run telemetry --tags cleanup \
    -e Delete_sinks_volume=true
  ./omnia.sh --run orchestrator --tags cleanup \
    -e cleanup_slurm=true -e cleanup_k8s=true
  ./omnia.sh --run discovery --tags cleanup
  ./omnia.sh --run image_build_manager --tags cleanup
  ./omnia.sh --run repo_manager --tags cleanup
  ./omnia.sh --run utils --tags cleanup

  ./omnia.sh --cleanup --all
)

Review each command's final Ansible recap as the sequence runs. Main full cleanup still performs its safety check and requests the exact confirmation yes; set -e stops the sequence when that safety check or any earlier command fails.

The following sections explain each command and its less-destructive options. For individual execution, choose the applicable command tab. Do not remove the shared virtual environment while another domain cleanup still needs it.

1. Clean up BuildStreaM¶

Skip this section when BuildStreaM was not deployed. The full cleanup removes the managed GitLab deployment, BuildStreaM service, watcher, automation artifacts, PostgreSQL service, NFS runtime directories, and BuildStreaM credentials. Ensure the configured GitLab host is reachable and its stored SSH credential is available before starting.

PostgreSQL data and volumes are preserved by default:

Using omnia.sh (recommended)Using ansible-playbook

Run on: OIM host

cd <OMNIA_SOURCE_PATH>/src/main
./omnia.sh --run build_stream --tags cleanup

Run on: OIM host

source /opt/omnia/activate-omnia.sh
cd <OMNIA_SOURCE_PATH>/src/build_stream/playbooks
ansible-playbook build_stream.yml --tags cleanup

To remove PostgreSQL data and volumes as part of a complete reset, run instead:

Using omnia.sh (recommended)Using ansible-playbook

Run on: OIM host

cd <OMNIA_SOURCE_PATH>/src/main
./omnia.sh --run build_stream --tags cleanup -e postgres_backup=false

Run on: OIM host

source /opt/omnia/activate-omnia.sh
cd <OMNIA_SOURCE_PATH>/src/build_stream/playbooks
ansible-playbook build_stream.yml --tags cleanup -e postgres_backup=false

To remove an individual BuildStreaM image group while retaining the deployed domain, do not run full domain cleanup. Use Clean up BuildStreaM image groups.

2. Clean up Telemetry¶

Run Telemetry cleanup while its Kubernetes cluster remains available. Full cleanup drains and removes all Telemetry sources before removing all sinks. Source-owned persistent volumes are deleted; Kafka, VictoriaMetrics, and VictoriaLogs volumes are preserved by default:

Using omnia.sh (recommended)Using ansible-playbook

Run on: OIM host

cd <OMNIA_SOURCE_PATH>/src/main
./omnia.sh --run telemetry --tags cleanup

Run on: OIM host

source /opt/omnia/activate-omnia.sh
cd <OMNIA_SOURCE_PATH>/src/telemetry/playbooks
ansible-playbook telemetry.yml --tags cleanup

Full cleanup removes the stored Telemetry credentials and logs by default. To retain both for a later deployment, run instead:

Using omnia.sh (recommended)Using ansible-playbook

Run on: OIM host

cd <OMNIA_SOURCE_PATH>/src/main
./omnia.sh --run telemetry --tags cleanup -e cleanup_credentials=false -e cleanup_logs=false

Run on: OIM host

source /opt/omnia/activate-omnia.sh
cd <OMNIA_SOURCE_PATH>/src/telemetry/playbooks
ansible-playbook telemetry.yml --tags cleanup -e cleanup_credentials=false -e cleanup_logs=false

Delete the preserved sink volumes only when a complete Telemetry data reset is intended:

Using omnia.sh (recommended)Using ansible-playbook

Run on: OIM host

cd <OMNIA_SOURCE_PATH>/src/main
./omnia.sh --run telemetry --tags cleanup -e Delete_sinks_volume=true

Run on: OIM host

source /opt/omnia/activate-omnia.sh
cd <OMNIA_SOURCE_PATH>/src/telemetry/playbooks
ansible-playbook telemetry.yml --tags cleanup -e Delete_sinks_volume=true

When Delete_sinks_volume=true is specified, Telemetry always deletes credentials, inputs, outputs and logs regardless of the cleanup_credentials and cleanup_logs values.

When retaining part of Telemetry, use selective sink cleanup or the applicable source cleanup operation instead of cleanup. See Clean Up Telemetry for dependency checks, source cleanup, credential and log preservation, and sink-volume deletion behavior.

3. Clean up Orchestrator¶

Back up required Slurm and Kubernetes shared data before this step. Preview the full cleanup plan without changing the environment:

Using omnia.sh (recommended)Using ansible-playbook

Run on: OIM host

cd <OMNIA_SOURCE_PATH>/src/main
DRY_RUN=true ./omnia.sh --run orchestrator --tags cleanup

Run on: OIM host

source /opt/omnia/activate-omnia.sh
cd <OMNIA_SOURCE_PATH>/src/orchestrator/playbooks
DRY_RUN=true ansible-playbook orchestrator.yml --tags cleanup

For an interactive run, use:

Using omnia.sh (recommended)Using ansible-playbook

Run on: OIM host

cd <OMNIA_SOURCE_PATH>/src/main
./omnia.sh --run orchestrator --tags cleanup

Run on: OIM host

source /opt/omnia/activate-omnia.sh
cd <OMNIA_SOURCE_PATH>/src/orchestrator/playbooks
ansible-playbook orchestrator.yml --tags cleanup

Orchestrator prompts independently before deleting Slurm and Kubernetes shared data. Only the exact response yes deletes the selected component's data. Any other response preserves its data, but cleanup still unmounts that storage and removes its /etc/fstab entry.

For a reviewed noninteractive decision, set both choices explicitly. For example, the following command deletes Slurm data and preserves Kubernetes data:

Using omnia.sh (recommended)Using ansible-playbook

Run on: OIM host

cd <OMNIA_SOURCE_PATH>/src/main
./omnia.sh --run orchestrator --tags cleanup \
  -e cleanup_slurm=true -e cleanup_k8s=false

Run on: OIM host

source /opt/omnia/activate-omnia.sh
cd <OMNIA_SOURCE_PATH>/src/orchestrator/playbooks
ansible-playbook orchestrator.yml --tags cleanup \
  -e cleanup_slurm=true -e cleanup_k8s=false

Full cleanup removes the Orchestrator credential file and Vault key by default. To preserve both credential artifacts while removing the other enabled components, run:

Using omnia.sh (recommended)Using ansible-playbook

Run on: OIM host

cd <OMNIA_SOURCE_PATH>/src/main
./omnia.sh --run orchestrator --tags cleanup \
  -e cleanup_credentials=false

Run on: OIM host

source /opt/omnia/activate-omnia.sh
cd <OMNIA_SOURCE_PATH>/src/orchestrator/playbooks
ansible-playbook orchestrator.yml --tags cleanup \
  -e cleanup_credentials=false

An explicit cleanup_credentials tag still removes the credential artifacts. To clean only one component, follow Clean up Orchestrator and use the standalone component flow with explicit component tags.

Danger

OpenCHAMI cleanup uninstalls the openchami and ochami RPMs and permanently deletes the boot-service-data, metadata-service-data, and postgres-data Podman volumes. Back up any required OpenCHAMI state before continuing.

Warning

SKIP_APPROVAL=true approves deletion of Slurm and Kubernetes shared data when cleanup_slurm or cleanup_k8s is omitted. Set both choices explicitly before using noninteractive cleanup.

4. Clean up Discovery¶

Discovery cleanup affects only the current project. It empties the project output directory and removes discovery_credentials.yml and its Vault key by default. Copy any discovered node mapping needed for later reuse before running:

Using omnia.sh (recommended)Using ansible-playbook

Run on: OIM host

cd <OMNIA_SOURCE_PATH>/src/main
./omnia.sh --run discovery --tags cleanup

Run on: OIM host

source /opt/omnia/activate-omnia.sh
cd <OMNIA_SOURCE_PATH>/src/discovery/playbooks
ansible-playbook discovery.yml --tags cleanup

To empty the current project's output while preserving its credentials, run:

Using omnia.sh (recommended)Using ansible-playbook

Run on: OIM host

cd <OMNIA_SOURCE_PATH>/src/main
./omnia.sh --run discovery --tags cleanup \
  -e cleanup_credentials=false

Run on: OIM host

source /opt/omnia/activate-omnia.sh
cd <OMNIA_SOURCE_PATH>/src/discovery/playbooks
ansible-playbook discovery.yml --tags cleanup \
  -e cleanup_credentials=false

Other Discovery input files and Discovery logs are preserved. For the credentials-only operation, see Clean up Discovery data.

5. Clean up Image Build Manager¶

Full Image Build Manager cleanup removes the locally managed MinIO service and data, the local registry and its data, build work directories, credentials, domain logs, and /root/.s3cfg when local MinIO is used. It also empties the shared Image Build Manager output and log roots, affecting every project that uses the configured domain data path:

Using omnia.sh (recommended)Using ansible-playbook

Run on: OIM host

cd <OMNIA_SOURCE_PATH>/src/main
./omnia.sh --run image_build_manager --tags cleanup

Run on: OIM host

source /opt/omnia/activate-omnia.sh
cd <OMNIA_SOURCE_PATH>/src/image_build_manager/playbooks
ansible-playbook image_build_manager.yml --tags cleanup

When PowerScale is the configured S3 provider, full cleanup preserves /root/.s3cfg and does not delete objects from PowerScale. If the goal is only to remove selected or all built images while retaining Image Build Manager services, use Clean up built images instead.

6. Clean up Repository Manager¶

Run Repository Manager cleanup after Image Build Manager no longer needs its package content. Full cleanup removes the Pulp service, container image, Pulp data, repository integration, credentials, and logs. Credentials and logs are removed by default without prompting:

Using omnia.sh (recommended)Using ansible-playbook

Run on: OIM host

cd <OMNIA_SOURCE_PATH>/src/main
./omnia.sh --run repo_manager --tags cleanup

Run on: OIM host

source /opt/omnia/activate-omnia.sh
cd <OMNIA_SOURCE_PATH>/src/repo_manager/playbooks
ansible-playbook repo_manager.yml --tags cleanup

To retain credentials and logs for a later deployment, run instead:

Using omnia.sh (recommended)Using ansible-playbook

Run on: OIM host

cd <OMNIA_SOURCE_PATH>/src/main
./omnia.sh --run repo_manager --tags cleanup \
  -e cleanup_credentials=false -e cleanup_logs=false

Run on: OIM host

source /opt/omnia/activate-omnia.sh
cd <OMNIA_SOURCE_PATH>/src/repo_manager/playbooks
ansible-playbook repo_manager.yml --tags cleanup \
  -e cleanup_credentials=false -e cleanup_logs=false

To remove selected repositories or artifacts while retaining Pulp, use Pulp cleanup instead of full Repository Manager cleanup.

7. Clean up Utils¶

The general Utils cleanup removes cluster-log artifacts, unattended-install temporary files and credentials, all OIM log-backup runs, and all Slurm configuration backup runs:

Using omnia.sh (recommended)Using ansible-playbook

Run on: OIM host

cd <OMNIA_SOURCE_PATH>/src/main
./omnia.sh --run utils --tags cleanup

Run on: OIM host

source /opt/omnia/activate-omnia.sh
cd <OMNIA_SOURCE_PATH>/src/utils/playbooks
export ANSIBLE_CONFIG=../ansible.cfg
ansible-playbook utils.yml --tags cleanup

The backup cleanup workflows do not apply retention or ask for confirmation. When any artifact or backup class must remain, follow Clean up Utils and run only the applicable scoped cleanup tags.

Remove the shared Omnia environment¶

After every required domain cleanup succeeds, choose one Main cleanup mode. For complete command behavior and verification, see Maintain the Main environment.

To remove the installed environment while preserving domain input, output, and log data under OMNIA_DATA_PATH, run:

Run on: OIM host

./omnia.sh --cleanup

To perform an explicitly approved full reset, run:

Run on: OIM host

./omnia.sh --cleanup --all

The full-reset safety check runs before the confirmation prompt. It refuses to continue when it finds an unsafe data path or deployed, generated, nonempty, or unrecognized domain state. No files are removed when this check fails. Run the matching domain cleanup or review and remove the reported retained path, then retry.

When the safety check passes, review the displayed paths and type exactly yes. The command removes the installed environment and all remaining data under OMNIA_DATA_PATH. Data in custom component roots or external storage is removed only by the applicable domain cleanup workflow.

Warning

Adding --skip-approval skips the Main confirmation prompt but does not bypass the full-reset safety checks.

Verification¶

Confirm that every selected domain cleanup has a successful Ansible recap.
Verify that the intended services, containers, Kubernetes resources, and mounts are absent.
Confirm that credentials, persistent volumes, shared data, and backups were preserved or removed according to the selected options.
After normal Main cleanup, confirm that the virtual environment and installed environment files are absent and OMNIA_DATA_PATH remains.
After --cleanup --all, confirm that the configured OMNIA_DATA_PATH is absent.

Next steps¶

Run ./omnia.sh --setup-venv to recreate the shared environment when required.
Initialize and deploy only the required domains in dependency order.
See Clean up Orchestrator for component-level Orchestrator cleanup.
See Pulp cleanup for selective repository cleanup.

Troubleshooting¶

A domain cleanup fails: Stop the sequence, correct the reported problem, and rerun that domain cleanup before removing its dependencies.
BuildStreaM cannot clean the GitLab host: Confirm that the configured GitLab host is reachable and that the BuildStreaM credential file still contains the GitLab SSH password.
Orchestrator shared data was preserved unexpectedly: Only the exact response yes approves interactive deletion. Rerun with reviewed cleanup_slurm and cleanup_k8s values when a noninteractive decision is required.
Main full cleanup reports incomplete domain cleanup: No files were removed. Run the cleanup tag for the reported domain. Preserve and then remove any reported status, log, output, or backup file that is intentionally retained before retrying.
A custom data path remains: Main full cleanup removes only OMNIA_DATA_PATH. Rerun the owning domain cleanup with the original project and component path configuration.
The virtual environment was removed too early: Set up the OIM again to restore the shared environment, complete the required domain cleanup, and then rerun the selected Main cleanup mode.