[OCPBUGS-104855]: Fix PPC example to use even CPU counts for HT compatibility

[kquinn1204]

Version(s):
4.20, 4.21, 4.22

Issue:
https://redhat.atlassian.net/browse/OCPBUGS-104855

Link to docs preview:

QE review:

• QE has approved this change.

Additional information:

The Performance Profile Creator (PPC) example command used --reserved-cpu-count=1 and --offlined-cpu-count=1, which fails on hyper-threading systems (the default) with:

Error: failed to make profile data from node handler: failed to compute the reserved and isolated CPUs: can't allocate odd number of CPUs from a NUMA Node

Affected modules:

• modules/cnf-running-the-performance-creator-profile.adoc (Podman procedure)
• modules/cnf-running-the-performance-creator-profile-offline.adoc (wrapper script procedure)

Changes (both modules):

1. Changes --reserved-cpu-count from 1 to 2 so the value is even and compatible with hyper-threading NUMA allocation.
2. Removes --offlined-cpu-count from the example entirely. On the existing 4-CPU example system, --reserved-cpu-count=2 plus --offlined-cpu-count=2 would consume all 4 CPUs, leaving zero for isolated workloads — an invalid configuration. The --offlined-cpu-count flag remains documented in the PPC argument reference table.
3. Updates the example output (log messages and generated PerformanceProfile YAML) to reflect the corrected CPU allocations: reserved: "0-1", isolated: 2-3.
4. Updates example output blocks from the old level=info msg="..." log format to the PPC: ... format introduced in v4.20. Verified by testing PPC images from v4.19 (old format) through v4.22 (new format) — the cutover occurred in v4.20.
5. (Wrapper script module only) Updates help output and --info invocation to reflect the v4.20+ subcommand syntax (info instead of --info=log).

Cherry-pick guidance for backports:

The branch has 3 commits:

• fc42ec6 — CPU count fix only (retains old level=info msg= format) — cherry-pick this for 4.19 and below
• 0936dc0 — Log format update for Podman module — relevant for 4.20+ only
• db62046 — Same fixes for wrapper script module (CPU counts + log format + info subcommand) — relevant for 4.20+ only; for 4.19 and below, the wrapper script module needs only the CPU count fix with old format

[openshift-ci-robot]

@kquinn1204: This pull request references Jira Issue OCPBUGS-104855, which is invalid:

• expected the bug to target the "5.0.0" version, but no target version was set

Comment /jira refresh to re-evaluate validity if changes to the Jira bug are made, or edit the title of this pull request to link to a different bug.

The bug has been updated to refer to the pull request using the external bug tracker.

Details

In response to this:

[OCPBUGS-104855]: Fix PPC example to use even CPU counts for HT compatibility. The Performance Profile Creator example used --reserved-cpu-count=1 and --offlined-cpu-count=1, which fails on hyper-threading systems because CPUs must be allocated in even numbers per NUMA node. Change reserved count to 2 and remove offlined count to keep the 4-CPU example consistent (2 reserved + 2 isolated = 4 total).

Version(s): 4.14 +

Issue:https://redhat.atlassian.net/browse/OCPBUGS-104855

Link to docs preview:

QE review:

• QE has approved this change.

Additional information:

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the openshift-eng/jira-lifecycle-plugin repository.

[ocpdocs-previewbot]

🤖 Thu Aug 06 15:38:16 - Prow CI generated the docs preview:

https://117426--ocpdocs-pr.netlify.app/openshift-enterprise/latest/scalability_and_performance/cnf-tuning-low-latency-nodes-with-perf-profile.html

[openshift-ci-robot]

@kquinn1204: No Jira issue is referenced in the title of this pull request.
To reference a jira issue, add 'XYZ-NNN:' to the title of this pull request and request another refresh with /jira refresh.

Details

In response to this:

Version(s):
4.14, 4.15, 4.16, 4.17, 4.18, 4.19, 4.20, 4.21, 4.22

Issue:
https://redhat.atlassian.net/browse/OCPBUGS-104855

Link to docs preview:

QE review:

• QE has approved this change.

Additional information:

The Performance Profile Creator (PPC) example command used --reserved-cpu-count=1 and --offlined-cpu-count=1, which fails on hyper-threading systems (the default) with:

Error: failed to make profile data from node handler: failed to compute the reserved and isolated CPUs: can't allocate odd number of CPUs from a NUMA Node

This fix:

1. Changes --reserved-cpu-count from 1 to 2 so the value is even and compatible with hyper-threading NUMA allocation.
2. Removes --offlined-cpu-count from the example entirely. On the existing 4-CPU example system, --reserved-cpu-count=2 plus --offlined-cpu-count=2 would consume all 4 CPUs, leaving zero for isolated workloads — an invalid configuration. The --offlined-cpu-count flag remains documented in the PPC argument reference table earlier in the module.
3. Updates the example output (log messages and generated PerformanceProfile YAML) to reflect the corrected CPU allocations: reserved: "0-1", isolated: 2-3.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the openshift-eng/jira-lifecycle-plugin repository.

[openshift-ci]

@kquinn1204: all tests passed!

Full PR test history. Your PR dashboard.

Details

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. I understand the commands that are listed here.