CNF-26546:tuned:irqs: set default_smp_affinity using irqaffinity=

[Tal-or]

When ovsDpdk CPUs are configured, add irqaffinity= to the kernel command line to set default_smp_affinity at boot time instead of manipulating it post-boot via clear-irqbalance-banned-cpus.sh.

AIA Human-AI blend, New content, Human-initiated, Reviewed, Claude Opus 4.6 v1.0

Summary by CodeRabbit

• New Features
  • Added IRQ affinity configuration to tuned profiles based on isolated, reserved, shared, and OVS-DPDK CPU settings.
• Bug Fixes
  • Stopped modifying the system’s default IRQ affinity mask while maintaining IRQBalance banned-CPU configuration.
  • Updated CPU isolation behavior to align with the revised IRQ affinity handling.
• Tests
  • Updated rendered configuration expectations for the revised CPU and IRQ affinity behavior.

[coderabbitai]

Walkthrough

The change derives IRQ-affinity CPUs from isolated, reserved, and shared CPUs minus OVS-DPDK CPUs, passes them to tuned templates for conditional kernel arguments, and removes default IRQ affinity mask modification from the irqbalance cleanup script.

Changes

IRQ affinity configuration

Layer / File(s)	Summary
Derive and render IRQ affinity pkg/performanceprofile/controller/performanceprofile/components/tuned/tuned.go, assets/performanceprofile/tuned/openshift-node-performance, assets/performanceprofile/tuned/openshift-node-performance-intel-x86	NewNodePerformance computes the IRQ-affinity CPU set and tuned profiles conditionally render the corresponding bootloader argument.
Preserve default affinity mask assets/performanceprofile/scripts/clear-irqbalance-banned-cpus.sh, pkg/performanceprofile/controller/performanceprofile/components/assets_scripts_test.go	The cleanup script no longer writes /proc/irq/default_smp_affinity, and tests for that removed behavior are deleted.
Refresh rendered outputs test/e2e/performanceprofile/testdata/render-expected-output/**	Expected MachineConfig and Tuned outputs reflect the shortened scripts, new IRQ-affinity placeholders, and embedded profile formatting updates.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Sequence Diagram(s)

sequenceDiagram
  participant NewNodePerformance
  participant TunedTemplate
  participant Bootloader
  NewNodePerformance->>TunedTemplate: Pass derived IRQ-affinity CPU set
  TunedTemplate->>Bootloader: Render cmdline_irqaffinity kernel argument

Loading

Suggested reviewers: swatisehgal, jmencak

🚥 Pre-merge checks | ✅ 14 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Test Structure And Quality	⚠️ Warning	Tests use bare Expect(...).ToNot(HaveOccurred()) throughout, and writeTempFile ignores its content arg, so scenarios aren't properly isolated.	Add descriptive failure messages to assertions and fix writeTempFile to write the provided content; consider DeferCleanup/AfterEach for temp files.

✅ Passed checks (14 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Stable And Deterministic Test Names	✅ Passed	The edited Go test file keeps static Ginkgo titles; no dynamic names, timestamps, IDs, or interpolated values were added.
Microshift Test Compatibility	✅ Passed	No new Ginkgo e2e tests were added in the diff; the changes are fixture YAMLs only and show no MicroShift-incompatible APIs/features.
Single Node Openshift (Sno) Test Compatibility	✅ Passed	No new Ginkgo e2e tests were added; changes are templates/scripts and unit-test removals, so no new SNO-specific risk is introduced.
Topology-Aware Scheduling Compatibility	✅ Passed	Changes only add boot-time irqaffinity tuning and remove a post-boot affinity script path; no pod scheduling, replicas, affinity, or topology logic was introduced.
Ote Binary Stdout Contract	✅ Passed	No changed entrypoint code writes to stdout; the only prints are GinkgoWriter inside It blocks, which are allowed.
Ipv6 And Disconnected Network Test Compatibility	✅ Passed	No new e2e Ginkgo tests were added; the PR only changes render-expected-output fixtures, so there are no new IPv4 or connectivity assumptions to flag.
No-Weak-Crypto	✅ Passed	Changed files only adjust performanceprofile scripts/tuned YAML; scans of diffs and decoded payloads found no weak-crypto or secret-comparison usage.
Container-Privileges	✅ Passed	Changed files are test fixtures only, and none contain privileged/hostPID/hostNetwork/hostIPC/allowPrivilegeEscalation/SYS_ADMIN/root settings.
No-Sensitive-Data-In-Logs	✅ Passed	No added logs exposing passwords/tokens/PII were found; the only tracing is set -x and echoing CPU masks/paths in the shell script.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly describes setting default_smp_affinity through the irqaffinity kernel parameter, which matches the main pull request objective.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[qodo-for-rh-openshift]

PR Summary by Qodo

Set default IRQ affinity via kernel irqaffinity= when ovsDpdk CPUs are configured

✨ Enhancement 🐞 Bug fix ⚙️ Configuration changes 🕐 20-40 Minutes

AI Description

• Add irqaffinity= kernel arg to set default_smp_affinity at boot for ovsDpdk setups.
• Stop post-boot mutation of /proc/irq/default_smp_affinity in clear-irqbalance-banned-cpus.sh.
• Compute irqaffinity CPU list as “all profile CPUs minus ovsDpdk CPUs” for tuned rendering.

Diagram

graph TD
  A["PerformanceProfile (CPU spec)"] --> B["tuned.go (compute IrqAffinityCpus)"] --> C["Tuned asset: openshift-node-performance"] --> D["Kernel cmdline: irqaffinity="] --> E["/proc/irq/default_smp_affinity"]
  C --> F["Script: clear-irqbalance-banned-cpus.sh"] --> G["/etc/sysconfig/irqbalance"]

Loading

High-Level Assessment

The following are alternative approaches to this PR:

1. Early-boot systemd unit to write /proc/irq/default_smp_affinity

• ➕ Avoids changing kernel cmdline (potentially simpler rollback).
• ➕ Can implement richer logic (e.g., topology-aware masks) at runtime.
• ➖ Still a post-boot mutation path; ordering vs CRI-O/irqbalance can be fragile.
• ➖ More moving parts (unit ordering, retries) than a kernel boot arg.

2. Rely only on irqbalance banned CPUs (no default_smp_affinity changes)

• ➕ Minimal change surface; keeps behavior limited to irqbalance.
• ➖ Does not address consumers like CRI-O that derive behavior from default_smp_affinity.
• ➖ Can lead to inconsistent IRQ masks when other components “restore” or recompute settings.

3. Use a sysctl/tuned sysfs write instead of kernel cmdline

• ➕ Keeps control inside tuned rather than bootloader args.
• ➕ Potentially more portable across bootloader implementations.
• ➖ Still depends on runtime ordering; may be too late for early IRQ routing decisions.
• ➖ More susceptible to being overwritten by other agents.

Recommendation: The PR’s approach (setting default_smp_affinity via irqaffinity= at boot) is the most robust for ovsDpdk cases because it establishes the desired default before user space components (e.g., CRI-O restore flows) can observe or overwrite it. The main review focus should be verifying the computed CPU list semantics (“all profile CPUs minus ovsDpdk CPUs”) matches the intended IRQ routing policy, and that the kernel argument format produced by cpuset.String() is accepted in the target environments.

Files changed (3) +14 / -25

Enhancement (1) +10 / -0

tuned.goCompute IrqAffinityCpus for ovsDpdk and pass into tuned template +10/-0

Compute IrqAffinityCpus for ovsDpdk and pass into tuned template

• Adds a new template argument ('IrqAffinityCpus') when 'Spec.CPU.OvsDpdk' is set. The value is derived as the union of isolated (+ovsDpdk) and reserved CPUs, minus the ovsDpdk CPU set, and is used by the tuned asset to render 'irqaffinity='.

pkg/performanceprofile/controller/performanceprofile/components/tuned/tuned.go

Refactor (1) +0 / -25

clear-irqbalance-banned-cpus.shRemove post-boot default_smp_affinity rewrite from irqbalance helper script +0/-25

Remove post-boot default_smp_affinity rewrite from irqbalance helper script

• Deletes the logic that masked OVS-DPDK CPUs out of '/proc/irq/default_smp_affinity'. The script now only manages 'IRQBALANCE_BANNED_CPUS' and keeps CRI-O’s original-banned-cpus file consistent.

assets/performanceprofile/scripts/clear-irqbalance-banned-cpus.sh

Other (1) +4 / -0

openshift-node-performanceAdd conditional irqaffinity= kernel arg rendering to tuned profile +4/-0

Add conditional irqaffinity= kernel arg rendering to tuned profile

• Introduces a new 'cmdline_irqaffinity' stanza that appends 'irqaffinity={{.IrqAffinityCpus}}' when the template variable is present. This shifts default IRQ affinity setup to boot time.

assets/performanceprofile/tuned/openshift-node-performance

[coderabbitai]

🧹 Nitpick comments (1)

pkg/performanceprofile/controller/performanceprofile/components/tuned/tuned.go (1)

78-86: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win

Add regression coverage for the CPU-set contract.

Extend pkg/performanceprofile/controller/performanceprofile/components/tuned/tuned_test.go to cover reserved/isolated/OVS-DPDK overlap, canonical output, and invalid empty results.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In
`@pkg/performanceprofile/controller/performanceprofile/components/tuned/tuned.go`
around lines 78 - 86, Extend the tests for the tuned template CPU-set
construction around the reserved/isolated merge and OVS-DPDK subtraction,
covering overlapping inputs, canonical ordering of the resulting affinity
string, and cases where subtraction produces an invalid empty set. Use the
existing test helpers and assertions in tuned_test.go, anchoring coverage to the
templateIrqAffinityCpus behavior and preserving the current error contract.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Nitpick comments:
In
`@pkg/performanceprofile/controller/performanceprofile/components/tuned/tuned.go`:
- Around line 78-86: Extend the tests for the tuned template CPU-set
construction around the reserved/isolated merge and OVS-DPDK subtraction,
covering overlapping inputs, canonical ordering of the resulting affinity
string, and cases where subtraction produces an invalid empty set. Use the
existing test helpers and assertions in tuned_test.go, anchoring coverage to the
templateIrqAffinityCpus behavior and preserving the current error contract.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Repository: openshift/coderabbit/.coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 5148f8fc-6698-4f5a-875b-80fa874cf6bc

📥 Commits

Reviewing files that changed from the base of the PR and between a9d25d5 and ef6d67e.

📒 Files selected for processing (3)

• assets/performanceprofile/scripts/clear-irqbalance-banned-cpus.sh
• assets/performanceprofile/tuned/openshift-node-performance
• pkg/performanceprofile/controller/performanceprofile/components/tuned/tuned.go

💤 Files with no reviewable changes (1)

• assets/performanceprofile/scripts/clear-irqbalance-banned-cpus.sh

[qodo-for-rh-openshift]

Code Review by Qodo

🐞 Bugs (0) 📘 Rule violations (0) 📎 Requirement gaps (0) 🎨 UX issues (0) 🔗 Cross-repo conflicts (0) 📜 Skill insights (0)

1. Stale SMP affinity tests ✓ Resolved 🐞 Bug ≡ Correctness

Description

clear-irqbalance-banned-cpus.sh no longer updates default_smp_affinity, but assets_scripts_test.go
still asserts the script clears OVS-DPDK CPUs from a default_smp_affinity file, so the unit tests
will fail.

Code

assets/performanceprofile/scripts/clear-irqbalance-banned-cpus.sh[L36-59]

-# CRI-O reads /proc/irq/default_smp_affinity to derive the IRQ banned mask
-# when pods with irq-load-balancing.crio.io=disable are scheduled.
-# If we don't remove the OVS-DPDK CPUs from default_smp_affinity here,
-# CRI-O will overwrite IRQBALANCE_BANNED_CPUS while ignoring the OVS-DPDK CPUs.
-SMP_AFFINITY="${3:-/proc/irq/default_smp_affinity}"
-if [ "${BANNED_CPUS}" != "${NONE}" ] && [ -f "${SMP_AFFINITY}" ]; then
-	# default_smp_affinity is comma-separated 32-bit hex groups (e.g. "ff,ffffffff,ffffffff")
-	IFS=',' read -ra smp < "${SMP_AFFINITY}"
-	n=${#smp[@]}
-	# pad BANNED_CPUS with leading zeros to match the same number of hex chars
-	padded="${BANNED_CPUS}"
-	while [ ${#padded} -lt $(( n * 8 )) ]; do
-		padded="0${padded}"
-	done
-	# clear banned bits from each 32-bit group: result = smp & ~banned
-	result=""
-	for (( i=0; i<n; i++ )); do
-		ban="${padded:$(( i * 8 )):8}"
-		val=$(printf "%08x" $(( 0x${smp[$i]} & ~0x${ban} )))
-		result+="${result:+,}${val}"
-	done
-	echo "Setting default_smp_affinity to ${result} (removing OVS-DPDK CPUs ${BANNED_CPUS} from default_smp_affinity mask)"
-	echo "${result}" > "${SMP_AFFINITY}"
-fi

Relevance

●●● Strong

Team previously updated assets_scripts_test.go when changing clear-irqbalance-banned-cpus.sh
behavior (PR #1546).

PR-#1546

ⓘ Recommendations generated based on similar findings in past PRs

Evidence

The script no longer contains any logic to read/write a default_smp_affinity path, while the unit
tests still pass a smpAffinity file path and assert its contents equal the cleared mask.

assets/performanceprofile/scripts/clear-irqbalance-banned-cpus.sh[1-34]
pkg/performanceprofile/controller/performanceprofile/components/assets_scripts_test.go[131-201]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
`clear-irqbalance-banned-cpus.sh` no longer edits `/proc/irq/default_smp_affinity`, but the unit tests still expect it to clear OVS-DPDK CPUs from a supplied `default_smp_affinity` file path.

## Issue Context
The script now only updates `IRQBALANCE_BANNED_CPUS` and the CRI-O backup file, leaving `default_smp_affinity` untouched. The existing tests invoke the script with a `default_smp_affinity` temp file and assert its contents change.

## Fix Focus Areas
- pkg/performanceprofile/controller/performanceprofile/components/assets_scripts_test.go[131-201]
- assets/performanceprofile/scripts/clear-irqbalance-banned-cpus.sh[1-34]

## Expected fix
- Either remove/replace the `DescribeTable("should clear ovs-dpdk cpus from default_smp_affinity", ...)` assertions to match the new script behavior, **or** reintroduce the old `default_smp_affinity` mutation if it is still required.
- If the new intended behavior is “set via kernel cmdline irqaffinity”, add a new test that validates the rendered tuned profile includes `irqaffinity=` when OVS-DPDK CPUs are configured (instead of testing the script modifies `default_smp_affinity`).

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

2. Shared CPUs omitted ✓ Resolved 🐞 Bug ≡ Correctness

Description

NewNodePerformance computes IrqAffinityCpus from reserved/isolated CPUs (minus ovsDpdk), ignoring
spec.cpu.shared; when MixedCpus is enabled, shared CPUs are included in kubelet ReservedSystemCPUs
but will be excluded from irqaffinity.

Code

pkg/performanceprofile/controller/performanceprofile/components/tuned/tuned.go[R78-86]

+			allCpus := isolatedSet
+			if profile.Spec.CPU.Reserved != nil {
+				reservedSet, err := cpuset.Parse(string(*profile.Spec.CPU.Reserved))
+				if err != nil {
+					return nil, fmt.Errorf("cannot parse reserved cpuset: %v", err)
+				}
+				allCpus = allCpus.Union(reservedSet)
+			}
+			templateArgs[templateIrqAffinityCpus] = allCpus.Difference(ovsDpdkSet).String()

Relevance

●● Moderate

No prior reviews found about including spec.cpu.shared in irqaffinity/MixedCpus calculations;
acceptance unclear.

ⓘ Recommendations generated based on similar findings in past PRs

Evidence

The new irqaffinity set is built only from isolated/reserved, while kubelet config explicitly adds
shared CPUs into ReservedSystemCPUs under MixedCPUs, demonstrating an inconsistency introduced by
the new computation.

pkg/performanceprofile/controller/performanceprofile/components/tuned/tuned.go[66-87]
pkg/performanceprofile/controller/performanceprofile/components/kubeletconfig/kubeletconfig.go[160-183]
pkg/apis/performanceprofile/v2/performanceprofile_types.go[118-152]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
When OVS-DPDK CPUs are configured, `IrqAffinityCpus` is computed as `(isolated ∪ reserved) \ ovsDpdk`. This ignores `spec.cpu.shared`, which is used under MixedCpus and is treated as part of the kubelet’s `ReservedSystemCPUs`.

## Issue Context
In MixedCpus mode, the operator unions `shared` CPUs into `ReservedSystemCPUs`, indicating they’re intended to behave like system/housekeeping CPUs. Excluding them from `irqaffinity=` prevents default IRQ affinity from including CPUs the kubelet is explicitly reserving for system use.

## Fix Focus Areas
- pkg/performanceprofile/controller/performanceprofile/components/tuned/tuned.go[66-91]
- pkg/performanceprofile/controller/performanceprofile/components/kubeletconfig/kubeletconfig.go[160-183]

## Expected fix
- When calculating `allCpus` for `IrqAffinityCpus`, also union in `spec.cpu.shared` **when MixedCpus is enabled** (e.g., via `profilecomponent.IsMixedCPUsEnabled(profile)`), then subtract `ovsDpdkSet`.
- Keep validation/error handling consistent with existing cpuset parsing patterns.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

3. Intel cmdline key missing ✓ Resolved 🐞 Bug ⚙ Maintainability

Description

openshift-node-performance introduces cmdline_irqaffinity, but
openshift-node-performance-intel-x86’s documented cmdline_* ordering placeholder block does not
include cmdline_irqaffinity, undermining Intel’s stable kernel-arg ordering mechanism.

Code

assets/performanceprofile/tuned/openshift-node-performance[R180-182]

+{{if .IrqAffinityCpus}}
+cmdline_irqaffinity=+irqaffinity={{.IrqAffinityCpus}}
+{{end}}

Relevance

●● Moderate

No historical evidence found; file path wasn’t present on default branch, so team precedent is
unclear.

ⓘ Recommendations generated based on similar findings in past PRs

Evidence

The base tuned profile adds a new cmdline_* variable and notes Intel ordering depends on these
names, while the Intel x86 ordering list does not include the new cmdline key.

assets/performanceprofile/tuned/openshift-node-performance[141-186]
assets/performanceprofile/tuned/openshift-node-performance-intel-x86[4-23]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
Intel x86 profile has an explicit list of `cmdline_*` placeholders to control/lock kernel-arg ordering, but the newly added `cmdline_irqaffinity` key is not declared there.

## Issue Context
The base profile explicitly states Intel relies on these names/order. Without adding the new key to the Intel placeholder list, `irqaffinity=` may be appended in a different position than intended on Intel nodes.

## Fix Focus Areas
- assets/performanceprofile/tuned/openshift-node-performance[141-186]
- assets/performanceprofile/tuned/openshift-node-performance-intel-x86[4-23]

## Expected fix
- Add `cmdline_irqaffinity=` to the Intel x86 placeholder block near the other cmdline_* entries so ordering remains deterministic when the base profile sets `cmdline_irqaffinity`.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

_{To customize comments, go to the Qodo configuration screen, or learn more in the docs.}

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In
`@test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/no-mcp/openshift-bootstrap-master_machineconfig.yaml`:
- Line 38: Remove or explicitly gate the unconditional set -x in the
clear-irqbalance-banned-cpus.sh source asset, then regenerate the embedded
payloads in all five listed MachineConfig fixtures:
test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/no-mcp/openshift-bootstrap-master_machineconfig.yaml:38-38,
test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/no-mcp/openshift-bootstrap-worker_machineconfig.yaml:38-38,
test/e2e/performanceprofile/testdata/render-expected-output/default/arm/manual_machineconfig.yaml:37-37,
test/e2e/performanceprofile/testdata/render-expected-output/default/cpuFrequency/manual_machineconfig.yaml:37-37,
and
test/e2e/performanceprofile/testdata/render-expected-output/default/manual_machineconfig.yaml:38-38,
ensuring none enables boot-time shell tracing by default.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Repository: openshift/coderabbit/.coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 740993a2-405e-4348-b0f9-7ca44ecc5b74

📥 Commits

Reviewing files that changed from the base of the PR and between a810671 and 293da86.

📒 Files selected for processing (18)

• test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/extra-mcp/openshift-bootstrap-master_machineconfig.yaml
• test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/extra-mcp/openshift-bootstrap-master_tuned.yaml
• test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/extra-mcp/openshift-bootstrap-worker_machineconfig.yaml
• test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/extra-mcp/openshift-bootstrap-worker_tuned.yaml
• test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/no-mcp/openshift-bootstrap-master_machineconfig.yaml
• test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/no-mcp/openshift-bootstrap-master_tuned.yaml
• test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/no-mcp/openshift-bootstrap-worker_machineconfig.yaml
• test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/no-mcp/openshift-bootstrap-worker_tuned.yaml
• test/e2e/performanceprofile/testdata/render-expected-output/default/arm/manual_machineconfig.yaml
• test/e2e/performanceprofile/testdata/render-expected-output/default/arm/manual_tuned.yaml
• test/e2e/performanceprofile/testdata/render-expected-output/default/cpuFrequency/manual_machineconfig.yaml
• test/e2e/performanceprofile/testdata/render-expected-output/default/cpuFrequency/manual_tuned.yaml
• test/e2e/performanceprofile/testdata/render-expected-output/default/manual_machineconfig.yaml
• test/e2e/performanceprofile/testdata/render-expected-output/default/manual_tuned.yaml
• test/e2e/performanceprofile/testdata/render-expected-output/default/pp-norps/manual_machineconfig.yaml
• test/e2e/performanceprofile/testdata/render-expected-output/default/pp-norps/manual_tuned.yaml
• test/e2e/performanceprofile/testdata/render-expected-output/no-ref/manual_machineconfig.yaml
• test/e2e/performanceprofile/testdata/render-expected-output/no-ref/manual_tuned.yaml

[coderabbitai]

📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win

Remove set -x from the embedded early-boot script.

The same updated clear-irqbalance-banned-cpus.sh payload enables shell tracing in every affected MachineConfig, causing boot-time command and configuration values to be written to the system journal. Remove or explicitly gate tracing in the source asset, then regenerate all rendered fixtures.

• test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/no-mcp/openshift-bootstrap-master_machineconfig.yaml#L38-L38: regenerate the payload without unconditional set -x.
• test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/no-mcp/openshift-bootstrap-worker_machineconfig.yaml#L38-L38: regenerate the payload without unconditional set -x.
• test/e2e/performanceprofile/testdata/render-expected-output/default/arm/manual_machineconfig.yaml#L37-L37: regenerate the payload without unconditional set -x.
• test/e2e/performanceprofile/testdata/render-expected-output/default/cpuFrequency/manual_machineconfig.yaml#L37-L37: regenerate the payload without unconditional set -x.
• test/e2e/performanceprofile/testdata/render-expected-output/default/manual_machineconfig.yaml#L38-L38: regenerate the payload without unconditional set -x.

📍 Affects 5 files

• test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/no-mcp/openshift-bootstrap-master_machineconfig.yaml#L38-L38 (this comment)
• test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/no-mcp/openshift-bootstrap-worker_machineconfig.yaml#L38-L38
• test/e2e/performanceprofile/testdata/render-expected-output/default/arm/manual_machineconfig.yaml#L37-L37
• test/e2e/performanceprofile/testdata/render-expected-output/default/cpuFrequency/manual_machineconfig.yaml#L37-L37
• test/e2e/performanceprofile/testdata/render-expected-output/default/manual_machineconfig.yaml#L38-L38

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In
`@test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/no-mcp/openshift-bootstrap-master_machineconfig.yaml`
at line 38, Remove or explicitly gate the unconditional set -x in the
clear-irqbalance-banned-cpus.sh source asset, then regenerate the embedded
payloads in all five listed MachineConfig fixtures:
test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/no-mcp/openshift-bootstrap-master_machineconfig.yaml:38-38,
test/e2e/performanceprofile/testdata/render-expected-output/bootstrap/no-mcp/openshift-bootstrap-worker_machineconfig.yaml:38-38,
test/e2e/performanceprofile/testdata/render-expected-output/default/arm/manual_machineconfig.yaml:37-37,
test/e2e/performanceprofile/testdata/render-expected-output/default/cpuFrequency/manual_machineconfig.yaml:37-37,
and
test/e2e/performanceprofile/testdata/render-expected-output/default/manual_machineconfig.yaml:38-38,
ensuring none enables boot-time shell tracing by default.

[Tal-or]

/retest

[Tal-or]

/retest

[Tal-or]

/test e2e-hypershift-pao

[qodo-for-rh-openshift]

PR-Agent: could not find a component named e2e-hypershift-pao in a supported language in this PR.

[yanirq]

@Tal-or are the rest of the script sections still relevant then ?

[Tal-or]

yes.

[yanirq]

you kept this part in the clear script IINM

[Tal-or]

yes, because the rest of logic is related to the dynamic irqbalancing, which is a pre-existing feature

[MarSik]

/lgtm

[openshift-ci]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: MarSik, Tal-or

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Details

Needs approval from an approver in each of these files:

• OWNERS [MarSik]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[openshift-merge-bot]

Scheduling required tests:
/test e2e-aws-operator
/test e2e-aws-ovn
/test e2e-aws-ovn-techpreview
/test e2e-gcp-pao
/test e2e-gcp-pao-updating-profile
/test e2e-gcp-pao-workloadhints
/test e2e-hypershift
/test e2e-hypershift-pao
/test e2e-no-cluster
/test e2e-upgrade

[Tal-or]

/retest

[openshift-ci-robot]

@Tal-or: This pull request references CNF-26546 which is a valid jira issue.

Warning: The referenced jira issue has an invalid target version for the target branch this PR targets: expected the story to target the "5.0.0" version, but no target version was set.

Details

In response to this:

When ovsDpdk CPUs are configured, add irqaffinity= to the kernel command line to set default_smp_affinity at boot time instead of manipulating it post-boot via clear-irqbalance-banned-cpus.sh.

AIA Human-AI blend, New content, Human-initiated, Reviewed, Claude Opus 4.6 v1.0

Summary by CodeRabbit

• New Features
• Added IRQ affinity configuration to tuned profiles based on isolated, reserved, shared, and OVS-DPDK CPU settings.
• Bug Fixes
• Stopped modifying the system’s default IRQ affinity mask while maintaining IRQBalance banned-CPU configuration.
• Updated CPU isolation behavior to align with the revised IRQ affinity handling.
• Tests
• Updated rendered configuration expectations for the revised CPU and IRQ affinity behavior.

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]

@Tal-or: 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.