Skip to content

docs/fix/test: NFS cleanup docs, telemetry Pulp OS version, and RFC 5737 test addresses - #5816

Merged
abhishek-sa1 merged 3 commits into
dell:stagingfrom
mithileshreddy04:staging
Nov 16, 2026
Merged

abhishek-sa1 merged 3 commits into
dell:stagingfrom
mithileshreddy04:staging

Conversation

@mithileshreddy04

Copy link
Copy Markdown
Collaborator

Description of the Solution

This PR bundles three small changes:

  1. The orchestrator cleanup README stated that Slurm and Kubernetes cleanup can delete NFS server data. That behaviour is not reachable (cleanup_nfs_server is fixed to false), and the runtime banners already say the data must be removed manually. The README now matches that.
  2. Telemetry derived pulp_os_version from the first inventory group containing a RHEL version. In hybrid clusters that group can belong to another domain, so Strimzi and Victoria downloads requested the wrong /rhel/<version>/ path and failed with HTTP 404. The derivation now considers only service_kube_* groups.
  3. Three example and test IP addresses in the telemetry tests were in the 100.x.x.x range, which the PR policy blocks. They are replaced with RFC 5737 documentation addresses (192.0.2.0/24).

Related Issue

  • Fixes #

Changes

Orchestrator cleanup docs (src/orchestrator/playbooks/cleanup/README.md)

  • States that Slurm and Kubernetes cleanup do not delete NFS server data, and that users remove it manually.
  • Removes the instructions for enabling server-side deletion, and the descriptions of its behaviour.
  • Removes the "By default" wording, which implied a user-facing setting.

Telemetry role (src/telemetry/roles/common/tasks/load_telemetry_config.yml)

  • Derives pulp_os_version only from service_kube_* inventory groups.
  • Renames the task to "Derive OS version from service K8s inventory group names".

Telemetry tests (test/telemetry/test_config.yml, test/telemetry/ut/test_idrac_bmc_inventory.py)

  • Replaces 100.10.x.x example and test addresses with 192.0.2.x documentation addresses. The test uses them only as keys.

Files Changed

File Change Type Description
src/orchestrator/playbooks/cleanup/README.md Modified Documents that NFS server data is not deleted by cleanup and must be removed manually.
src/telemetry/roles/common/tasks/load_telemetry_config.yml Modified Restricts the RHEL version derivation to service_kube_* groups.
test/telemetry/test_config.yml Modified Replaces a 100.x example address with 192.0.2.61.
test/telemetry/ut/test_idrac_bmc_inventory.py Modified Replaces 100.x test addresses with 192.0.2.120 and 192.0.2.121.

Testing

  • Not run: Python and Ansible are not available in the development shell, so no tests or playbook checks were executed.
  • Required before merge:
    • pytest ut/test_idrac_bmc_inventory.py -v from test/telemetry.
    • ansible-playbook syntax check and a telemetry deploy against a hybrid inventory where a non-K8s group uses a lower RHEL version. Confirm the Strimzi download requests /rhel/10.2/.
    • Confirm the PR policy check passes for these files.

Backward Compatibility

  • No behaviour changes. The README change describes existing behaviour; the server-side deletion code path is unreachable.
  • The telemetry change alters only which inventory groups supply the RHEL version. Inventories with a service_kube_* group are unaffected. If no such group exists, the existing fail task still reports the error.
  • The address change affects only test fixtures and an example comment.

…es and tests

The PR policy blocks 100.x.x.x IPv4 addresses in newly added lines. Three
example and test addresses in the telemetry tests used that range:
test_config.yml (idrac_nft_bmc_ip example) and test_idrac_bmc_inventory.py
(BMC IPs used as test keys).

Replace them with 192.0.2.0/24 documentation addresses (RFC 5737). The test
logic is unchanged: the addresses are only keys, and the function under test
does not check address ranges.

Signed-off-by: Mithilesh Reddy <mithilesh.reddy@dell.com>
The telemetry role derived pulp_os_version from the first orchestrator
inventory group whose name contains a RHEL version. In hybrid clusters
that group can belong to another domain (for example a login or slurm
node on RHEL 10.0), so the version resolved to 10.0 while the telemetry
packages are synced for 10.2. Strimzi and Victoria Metrics downloads
then failed with HTTP 404.

Telemetry runs only on the service K8s cluster, so consider only
service_kube_* groups when deriving the version.

Signed-off-by: Mithilesh Reddy <mithilesh.reddy@dell.com>
…NFS server data

The cleanup README described server-side NFS deletion as a setting users
could enable, and said it was controlled by cleanup_nfs_server. That flag
is internal and fixed to false, and the runtime banners already report that
NFS share data was not automatically cleaned up and must be removed manually.

Align the README with that behaviour: cleanup unmounts the component storage
and removes its fstab entries, and NFS server data is removed manually. Remove
the instructions for enabling server-side deletion and the descriptions of its
behaviour, and drop the "By default" wording that implied a setting.

Signed-off-by: Mithilesh Reddy <mithilesh.reddy@dell.com>
@abhishek-sa1
abhishek-sa1 merged commit e1316fd into dell:staging Nov 16, 2026
16 of 18 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants