Skip to content

Latest commit

 

History

History
175 lines (108 loc) · 6.18 KB

CLEANUP.md

File metadata and controls

175 lines (108 loc) · 6.18 KB

Cleaning up the Environment

In the following we describe which objects are created during the image build process on the build system, the NFS Server and in the Red Hat® OpenShift® Container Platform and how they can be removed.

Contents

Table of Contents

Cleaning up the Build LPAR

Cleaning up the Temporary Image Build Files

During the image build tools/image-build temporary files are copied from the reference SAP system hosts to the build system. These temporary files are stored in directory <temp-root>. If you do not explicitly specify a temporary directory as argument when building the images, /data/tmp is used.

It may be required to clean up temporary build data

rm -rf /data/tmp/soos-build-<flavor>.<temp-uuid>*

Cleaning up the Local Image Repository

To remove one or more of the previously created images from the local podman repository on your build LPAR:

  • list the images by running
  # podman images
  • copy the image ID and remove the image by issuing:
  # podman rmi -f <image-id>

Cleaning up the NFS Server

Cleaning up Orphaned Overlay Filesystems

These tasks are normally executed out of tools/ocp-deployment --remove or tools/nfs-overlay-teardown

⚠️ Before any manual clean up action on the NFS server ensure that the deployment in scope is stopped - so that the direcories are not mounted any more via NFS in a pod.

  • Edit the NFS export list /etc/exports on the NFS server and remove all orphaned exported filesystems

  • execute exportfs -a as root user on the NFS server to stop exporting

  • unmount all unnecessary overlay filesystems

  • Edit /etc/fstab on the NFS server and remove all the unneccesary overlay filesystems

Cleaning up the base directory

Once all deployments are deleted and there is no intention to reactivate them once again then the initial copy of {data,log} for the SAP HANA system can be removed too.

Before this action see Cleaning up the Orphaned Overlay Filesystems to ensure that no overlay filesystems is present any more.

Identify the base directory nfs.bases.copy in the config file config.yaml

Example:

bases:
    description: Base directories
    copy:
      description: Path where directories {data,log} of the original
                   SAP HANA system are copied to
      required: true
      value: '/nfsexp/soos/data'

and remove the SAP HANA data in it.

Cleaning up the Red Hat OpenShift Container Platform

Deleting Unnecessary OC Projects

  • Login as OC cluster user, e.g. by running

    tools/oc-login -u

    on the build LPAR.

  • Delete the project

    oc delete project <ocp-project-name>

    This will delete the project with all objects in it.

Deleting Old Images from OCP Cluster Registry

Cluster administrators can periodically prune older versions of images that are no longer in use, but are still taking up disk space.

  • Login as OC cluster admin, e.g. by running

    tools/oc-login -a

    on the build LPAR.

  • Execute the image pruning via

    oc adm prune images --confirm --registry-url=https://default-route-openshift-image-registry.apps.<ocp-domain-name>

Cleaning up the Red Had OpenShift Container Platform Helper Node

Deleting orphaned entries in the HAProxy configuration

Check the connections using the web monitor for HAProxy. On the Red Had OpenShift Container Platform helper node the web monitor can be reached via a browser using http://<ocp-helper-node>:9000

If all connections of one deployment <app-name> are down, the deployment is probably stopped or even removed.

HAProxy cleanup

  • As long as the deployment <app-name> is not removed from the RedHat OpenShift Cluster, the HAProxy entries of it can be removed by executing

    tools/ocp-haproxy-forwarding --remove --app-name=<app-name>

    on the build LPAR.

  • If the deployment <app-name> is already removed from the RedHat OpenShift Cluster, then the it cannot be removed from the HAProxy configuration using the tools.

    Logon as user root to the Red Had OpenShift Container Platform helper node, and

    • remove the orphaned entries by editing /etc/haproxy/haproxy.cfg

    • Afterwards reload the HAProxy service

      systemctl reload haproxy