Add nwswitch/link validations for host and VM side#23
Open
bogdando wants to merge 1 commit into
Open
Conversation
|
Note Reviews pausedIt looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the Use the following commands to manage reviews:
Use the checkboxes below for quick actions:
📝 WalkthroughWalkthroughAdds an Ansible NVSwitch validation playbook with configurable defaults, mode-specific host checks for FABRIC_MODE 0–2, and in-VM GPU isolation checks covering visibility, NVLink topology, fabric state, partitions, and mdev instances. ChangesNVSwitch validation
Estimated code review effort: 4 (Complex) | ~45 minutes Suggested reviewers: Sequence Diagram(s)sequenceDiagram
participant nvswitch_validation
participant edpm_hosts
participant gpu_vms
nvswitch_validation->>edpm_hosts: run mode-specific host collection
edpm_hosts->>edpm_hosts: assert services, fabric state, partitions, or mdev count
nvswitch_validation->>gpu_vms: run mode-specific VM collection
gpu_vms->>gpu_vms: assert GPU count and NVLink topology isolation
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
bogdando
force-pushed
the
nvswitch
branch
2 times, most recently
from
July 2, 2026 17:05
531ad82 to
7467c4d
Compare
bogdando
marked this pull request as ready for review
July 10, 2026 15:50
Contributor
Author
|
confirmed it works with NVSwitch and Fabric Manager mode 0 on hgx b 300 guest VM |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 9
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (2)
gpu-validation/tasks/nvswitch_mode2_isolation.yaml (1)
33-41: 🎯 Functional Correctness | 🟠 Major | ⚡ Quick winUse a double-escaped word boundary here.
'\bNV[0-9]'is parsed by Jinja as a backspace character, so the search never matches NVLink lines and the Mode 2 assertion will fail.Proposed fix
- | select('search', '\bNV[0-9]') + | select('search', '\\bNV[0-9]')🤖 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 `@gpu-validation/tasks/nvswitch_mode2_isolation.yaml` around lines 33 - 41, Update the `nvswitch_vm_nvlink_lines` expression in the “Set in-VM topology facts” task to use a double-escaped word boundary (`'\\bNV[0-9]'`) so Jinja passes the regex boundary correctly and NVLink lines are matched.gpu-validation/tasks/nvswitch_mode0_isolation.yaml (1)
36-44: 🎯 Functional Correctness | 🟠 Major | ⚡ Quick winFix the NVLink regex in
gpu-validation/tasks/nvswitch_mode0_isolation.yaml:42.'\bNV[0-9]'becomes a backspace escape here, sonvswitch_vm_nvlink_linesstays empty and the multi-GPU assertion innvswitch_mode0_isolation_assertions.yamlwill always fail. Use'\\bNV[0-9]'to match the shared files.🤖 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 `@gpu-validation/tasks/nvswitch_mode0_isolation.yaml` around lines 36 - 44, Fix the regex used by the nvswitch_vm_nvlink_lines fact in the “Set in-VM topology facts” task: escape the backslash as '\\bNV[0-9]' so it performs a word-boundary NVLink match instead of interpreting \b as a backspace, consistent with the shared files.
🧹 Nitpick comments (2)
gpu-validation/tasks/nvswitch_mode0.yaml (1)
27-89: 📐 Maintainability & Code Quality | 🔵 Trivial | 🏗️ Heavy liftDuplicated collection logic across all three mode host files.
The FM service status collection+facts (Lines 27-46), Fabric State collection+facts (Lines 48-70), and NVLink topology collection+facts (Lines 72-89) here are copy-pasted almost verbatim in
nvswitch_mode1.yaml(Lines 4-84) andnvswitch_mode2.yaml(Lines 6-82), differing only in the FM service-list variable and minor debug text. Consider extracting a shared parameterized task file for the common parts, leaving each mode file to add only its mode-specific checks (cfg parsing,fmpm -l,mdevctl list).This also appears to overlap with the "Shared host collection" layer (
nvswitch.yaml), which per the PR stack is described as collecting "Fabric Manager state, GPU fabric state, activated partitions, and topology" — the same data collected here. Worth confirming these aren't run twice per host.🤖 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 `@gpu-validation/tasks/nvswitch_mode0.yaml` around lines 27 - 89, Extract the duplicated FM service, GPU fabric-state, and NVLink topology collection into a shared parameterized task file, referenced by nvswitch_mode0.yaml, nvswitch_mode1.yaml, and nvswitch_mode2.yaml with each mode’s service-list variable and mode-specific messaging as inputs. Reconcile this shared collection with the existing nvswitch.yaml host-collection layer and remove or reuse overlapping tasks so each host gathers these facts only once, while leaving mode-specific checks in the individual mode files.gpu-validation/tasks/nvswitch_mode1_isolation.yaml (1)
1-40: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick winExact duplicate of
gpu-validation/tasks/nvswitch_isolation.yaml. Consolidate this into the shared task file to avoid drift.🤖 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 `@gpu-validation/tasks/nvswitch_mode1_isolation.yaml` around lines 1 - 40, Remove the duplicated task definitions from nvswitch_mode1_isolation.yaml and reuse the shared nvswitch_isolation.yaml task file instead. Preserve the existing GPU count, topology collection, debug output, and fact-setting behavior through the shared include/import mechanism.
🤖 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 `@gpu-validation/tasks/nvswitch_assertions.yaml`:
- Around line 12-24: Add a separate assertion in the Fabric State validation
task to verify the detected GPU count matches
nvswitch_validation_expected_gpu_count, in addition to the existing
nvswitch_fabric_completed_count check. Update fail_msg and success_msg as needed
to clearly report the discovered count, using the variables established by the
GPU discovery tasks.
In `@gpu-validation/tasks/nvswitch_isolation_assertions.yaml`:
- Around line 31-42: Remove the redundant “Assert GPU count in topo matches
expected” assertion from nvswitch_isolation_assertions.yaml, since
nvswitch_vm_gpu_count comes from nvidia-smi query-gpu output rather than
_vm_topo. Apply the same removal to nvswitch_mode1_isolation_assertions.yaml and
nvswitch_mode2_isolation_assertions.yaml, including their associated messages.
In `@gpu-validation/tasks/nvswitch_mode0.yaml`:
- Around line 35-41: Fix the Jinja spacing in the nvswitch_fm_service_results
set_fact expression by formatting the zip(...) call according to Ansible-lint
spacing conventions, without changing its mapping or resulting values.
- Around line 21-25: Fix the regex backreference in the “Set FABRIC_MODE config
fact” task by changing the regex_search replacement argument from '\1' to '\\1',
preserving the existing | first pipeline and default handling.
In `@gpu-validation/tasks/nvswitch_mode1.yaml`:
- Around line 12-18: The Jinja expression assigning nvswitch_fm_service_results
has inconsistent spacing around the zip filter, matching the issue in
nvswitch_mode0.yaml. Reformat the map/zip pipeline with the project’s required
spacing conventions while preserving its behavior.
- Around line 1-84: Add a validation task to the mode 1 and mode 2 workflows,
matching the existing fabricmanager.cfg/FABRIC_MODE check in
nvswitch_mode0.yaml, and ensure the shared nvswitch.yaml flow also invokes it.
Reference the corresponding task or validation symbol from mode 0 to keep
expected-value handling and failure reporting consistent, so stale FABRIC_MODE
settings are detected before other checks run.
In `@gpu-validation/tasks/nvswitch_mode2.yaml`:
- Around line 14-20: The Jinja expression assigning nvswitch_fm_service_results
has inconsistent spacing around the zip filter, triggering the lint warning.
Update the expression in the “Set FM service status facts” task to match the
spacing convention already used in nvswitch_mode0.yaml and nvswitch_mode1.yaml,
without changing its behavior.
In `@nvswitch_validation.yaml`:
- Around line 36-40: Remove ignore_errors: true from both host-side and VM
assertion tasks in nvswitch_validation.yaml so failures from the gpu-validation
role propagate and cause the playbook to fail.
- Around line 31-34: Move the default definition of
nvswitch_validation_fabric_mode out of the gpu-validation role defaults and
define it in the play, inventory, or extra-vars instead. Ensure the include_role
task using tasks_from "nvswitch_mode{{ nvswitch_validation_fabric_mode }}.yaml"
resolves the selector from play context, including the corresponding occurrence
around the additional referenced block.
---
Outside diff comments:
In `@gpu-validation/tasks/nvswitch_mode0_isolation.yaml`:
- Around line 36-44: Fix the regex used by the nvswitch_vm_nvlink_lines fact in
the “Set in-VM topology facts” task: escape the backslash as '\\bNV[0-9]' so it
performs a word-boundary NVLink match instead of interpreting \b as a backspace,
consistent with the shared files.
In `@gpu-validation/tasks/nvswitch_mode2_isolation.yaml`:
- Around line 33-41: Update the `nvswitch_vm_nvlink_lines` expression in the
“Set in-VM topology facts” task to use a double-escaped word boundary
(`'\\bNV[0-9]'`) so Jinja passes the regex boundary correctly and NVLink lines
are matched.
---
Nitpick comments:
In `@gpu-validation/tasks/nvswitch_mode0.yaml`:
- Around line 27-89: Extract the duplicated FM service, GPU fabric-state, and
NVLink topology collection into a shared parameterized task file, referenced by
nvswitch_mode0.yaml, nvswitch_mode1.yaml, and nvswitch_mode2.yaml with each
mode’s service-list variable and mode-specific messaging as inputs. Reconcile
this shared collection with the existing nvswitch.yaml host-collection layer and
remove or reuse overlapping tasks so each host gathers these facts only once,
while leaving mode-specific checks in the individual mode files.
In `@gpu-validation/tasks/nvswitch_mode1_isolation.yaml`:
- Around line 1-40: Remove the duplicated task definitions from
nvswitch_mode1_isolation.yaml and reuse the shared nvswitch_isolation.yaml task
file instead. Preserve the existing GPU count, topology collection, debug
output, and fact-setting behavior through the shared include/import mechanism.
🪄 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: Organization UI
Review profile: CHILL
Plan: Enterprise
Run ID: c7fffd94-0a7d-49d6-bad6-e0ce6b431b7b
📒 Files selected for processing (18)
gpu-validation/defaults/main.yamlgpu-validation/tasks/nvswitch.yamlgpu-validation/tasks/nvswitch_assertions.yamlgpu-validation/tasks/nvswitch_isolation.yamlgpu-validation/tasks/nvswitch_isolation_assertions.yamlgpu-validation/tasks/nvswitch_mode0.yamlgpu-validation/tasks/nvswitch_mode0_assertions.yamlgpu-validation/tasks/nvswitch_mode0_isolation.yamlgpu-validation/tasks/nvswitch_mode0_isolation_assertions.yamlgpu-validation/tasks/nvswitch_mode1.yamlgpu-validation/tasks/nvswitch_mode1_assertions.yamlgpu-validation/tasks/nvswitch_mode1_isolation.yamlgpu-validation/tasks/nvswitch_mode1_isolation_assertions.yamlgpu-validation/tasks/nvswitch_mode2.yamlgpu-validation/tasks/nvswitch_mode2_assertions.yamlgpu-validation/tasks/nvswitch_mode2_isolation.yamlgpu-validation/tasks/nvswitch_mode2_isolation_assertions.yamlnvswitch_validation.yaml
There was a problem hiding this comment.
Actionable comments posted: 4
🤖 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 `@gpu-validation/tasks/nvswitch_mode0.yaml`:
- Around line 21-25: In the “Set FABRIC_MODE config fact” task, guard the result
of regex_search before applying first. Update the nvswitch_fabric_mode_cfg_value
expression to default a missing or malformed regex result to an empty list (or
equivalent fallback) before first, then retain -1 as the final fallback so
validation can report invalid configuration.
In `@nvswitch_validation.yaml`:
- Around line 26-33: Add a localhost preflight play before the existing
Host-side checks play that validates inventory membership with assertions:
require edpm_hosts to contain at least one host and gpu_vms to contain at least
two hosts. Use localhost for this validation so it always executes even when the
target groups are empty or unmatched, and fail with clear messages when either
requirement is not met.
- Around line 37-41: Update the assertion for nvswitch_validation_fabric_mode to
validate the original value against the exact allowed strings ('0', '1', and
'2') instead of relying on the int filter, then normalize it once to an integer
and reuse that normalized variable in all four task includes using tasks_from.
- Around line 4-9: Ensure both the edpm_hosts and gpu_vms plays in
nvswitch_validation.yaml use one shared source for
nvswitch_validation_fabric_mode, such as extra-vars or group_vars/all, rather
than resolving it independently; alternatively, add a pre-dispatch assertion
that both play contexts resolve the same value.
🪄 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: Organization UI
Review profile: CHILL
Plan: Enterprise
Run ID: ad1a483b-8bd8-4dec-a361-b906638eb54b
📒 Files selected for processing (18)
gpu-validation/defaults/main.yamlgpu-validation/tasks/nvswitch.yamlgpu-validation/tasks/nvswitch_assertions.yamlgpu-validation/tasks/nvswitch_isolation.yamlgpu-validation/tasks/nvswitch_isolation_assertions.yamlgpu-validation/tasks/nvswitch_mode0.yamlgpu-validation/tasks/nvswitch_mode0_assertions.yamlgpu-validation/tasks/nvswitch_mode0_isolation.yamlgpu-validation/tasks/nvswitch_mode0_isolation_assertions.yamlgpu-validation/tasks/nvswitch_mode1.yamlgpu-validation/tasks/nvswitch_mode1_assertions.yamlgpu-validation/tasks/nvswitch_mode1_isolation.yamlgpu-validation/tasks/nvswitch_mode1_isolation_assertions.yamlgpu-validation/tasks/nvswitch_mode2.yamlgpu-validation/tasks/nvswitch_mode2_assertions.yamlgpu-validation/tasks/nvswitch_mode2_isolation.yamlgpu-validation/tasks/nvswitch_mode2_isolation_assertions.yamlnvswitch_validation.yaml
🚧 Files skipped from review as they are similar to previous changes (13)
- gpu-validation/tasks/nvswitch_mode2_assertions.yaml
- gpu-validation/tasks/nvswitch_assertions.yaml
- gpu-validation/defaults/main.yaml
- gpu-validation/tasks/nvswitch_mode0_isolation_assertions.yaml
- gpu-validation/tasks/nvswitch_mode0_isolation.yaml
- gpu-validation/tasks/nvswitch_isolation_assertions.yaml
- gpu-validation/tasks/nvswitch_mode2_isolation_assertions.yaml
- gpu-validation/tasks/nvswitch_mode1_isolation.yaml
- gpu-validation/tasks/nvswitch_isolation.yaml
- gpu-validation/tasks/nvswitch_mode0_assertions.yaml
- gpu-validation/tasks/nvswitch_mode1_assertions.yaml
- gpu-validation/tasks/nvswitch_mode2_isolation.yaml
- gpu-validation/tasks/nvswitch_mode1_isolation_assertions.yaml
MiguelCarpio
requested changes
Jul 13, 2026
There was a problem hiding this comment.
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 `@nvswitch_validation.yaml`:
- Around line 60-70: Update the fabric-mode assertions for both the nvswitch and
gpu_vms plays to validate the raw value is numeric before converting it,
preventing non-numeric inputs from being coerced to 0. Preserve acceptance of
numeric values 0, 1, and 2 and ensure invalid strings trigger the existing
fail_msg before any include_role/tasks_from execution.
🪄 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: Organization UI
Review profile: CHILL
Plan: Enterprise
Run ID: e04bf955-9900-448e-baed-d7db04ea436f
📒 Files selected for processing (18)
gpu-validation/defaults/main.yamlgpu-validation/tasks/nvswitch.yamlgpu-validation/tasks/nvswitch_assertions.yamlgpu-validation/tasks/nvswitch_isolation.yamlgpu-validation/tasks/nvswitch_isolation_assertions.yamlgpu-validation/tasks/nvswitch_mode0.yamlgpu-validation/tasks/nvswitch_mode0_assertions.yamlgpu-validation/tasks/nvswitch_mode0_isolation.yamlgpu-validation/tasks/nvswitch_mode0_isolation_assertions.yamlgpu-validation/tasks/nvswitch_mode1.yamlgpu-validation/tasks/nvswitch_mode1_assertions.yamlgpu-validation/tasks/nvswitch_mode1_isolation.yamlgpu-validation/tasks/nvswitch_mode1_isolation_assertions.yamlgpu-validation/tasks/nvswitch_mode2.yamlgpu-validation/tasks/nvswitch_mode2_assertions.yamlgpu-validation/tasks/nvswitch_mode2_isolation.yamlgpu-validation/tasks/nvswitch_mode2_isolation_assertions.yamlnvswitch_validation.yaml
🚧 Files skipped from review as they are similar to previous changes (16)
- gpu-validation/defaults/main.yaml
- gpu-validation/tasks/nvswitch_mode1_assertions.yaml
- gpu-validation/tasks/nvswitch_assertions.yaml
- gpu-validation/tasks/nvswitch.yaml
- gpu-validation/tasks/nvswitch_mode1_isolation.yaml
- gpu-validation/tasks/nvswitch_mode2_isolation.yaml
- gpu-validation/tasks/nvswitch_mode0_isolation.yaml
- gpu-validation/tasks/nvswitch_isolation.yaml
- gpu-validation/tasks/nvswitch_mode2_assertions.yaml
- gpu-validation/tasks/nvswitch_mode0_isolation_assertions.yaml
- gpu-validation/tasks/nvswitch_isolation_assertions.yaml
- gpu-validation/tasks/nvswitch_mode0_assertions.yaml
- gpu-validation/tasks/nvswitch_mode1_isolation_assertions.yaml
- gpu-validation/tasks/nvswitch_mode0.yaml
- gpu-validation/tasks/nvswitch_mode2.yaml
- gpu-validation/tasks/nvswitch_mode1.yaml
There was a problem hiding this comment.
Actionable comments posted: 4
♻️ Duplicate comments (1)
gpu-validation/tasks/nvswitch_mode1.yaml (1)
1-84: 🎯 Functional Correctness | 🟠 Major | ⚡ Quick winFABRIC_MODE cfg check still missing for mode 1 (and mode 2).
A prior review flagged that
nvswitch_mode1.yaml/nvswitch_mode2.yamldon't validatefabricmanager.cfg'sFABRIC_MODEthe waynvswitch_mode0.yamldoes, leaving stale FM config undetected for these modes. That comment wasn't marked addressed and the gap is still present in the current file.🤖 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 `@gpu-validation/tasks/nvswitch_mode1.yaml` around lines 1 - 84, The mode 1 validation tasks do not inspect fabricmanager.cfg for the expected FABRIC_MODE. Add a configuration check to nvswitch_mode1.yaml, matching the established validation used by nvswitch_mode0.yaml, and fail or report when the configured value does not match the mode 1 expectation. Apply the corresponding check to nvswitch_mode2.yaml as well if it shares the same omission.
🧹 Nitpick comments (1)
gpu-validation/tasks/nvswitch.yaml (1)
25-84: 📐 Maintainability & Code Quality | 🔵 Trivial | 🏗️ Heavy liftFabric-state / fmpm / topology collection is duplicated across mode files.
This shared file's FM-service, fabric-state, fmpm-partition, and topology collection (lines 12-84) is copy-pasted almost verbatim into
nvswitch_mode1.yaml(same header comment, same commands), and its fabric-state + topology blocks are duplicated again innvswitch_mode0.yamlandnvswitch_mode2.yaml. If the mode dispatcher runs this shared file and then the mode-specific file for every run (as the PR stack description implies), the same privileged commands (nvidia-smi -q,fmpm -l,nvidia-smi topo -m) execute twice, and the mode-specific run overwritesgpu_validation_nvswitch_fabric_states/gpu_validation_nvswitch_topo_output/etc. Depending on task ordering relative tonvswitch_assertions.yaml, this duplication could also cause the shared assertions to evaluate against facts overwritten by the mode-specific run rather than the intended shared collection. Consider having the mode-specific files only add their unique checks (FABRIC_MODE cfg, mdev list, mode-specific FM services) and rely on this shared file for fabric state/topology.🤖 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 `@gpu-validation/tasks/nvswitch.yaml` around lines 25 - 84, The shared collection tasks should be the sole source for fabric state, fmpm partitions, and NVLink topology facts. Remove the duplicated collection blocks from nvswitch_mode0.yaml, nvswitch_mode1.yaml, and nvswitch_mode2.yaml, leaving those files to perform only their mode-specific checks and FM-service work; preserve the shared task symbols gpu_validation_nvswitch_fabric_states, gpu_validation_nvswitch_fmpm_output, and gpu_validation_nvswitch_topo_output.
🤖 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 `@gpu-validation/tasks/nvswitch_mode1_isolation_assertions.yaml`:
- Line 18: Update the NVLink isolation assertion using
gpu_validation_nvswitch_vm_nvlink_lines so it requires the expected minimum
number of connected GPU rows for the VM, rather than merely checking that at
least one row exists; preserve the existing topology validation flow and use the
established GPU-count or expected-row value if available.
In `@gpu-validation/tasks/nvswitch_mode2_isolation.yaml`:
- Line 39: Update the regex passed to the select search in the nvswitch
isolation task so the word-boundary escape is preserved through Jinja parsing,
using the same escaped-pattern convention as the later \d expression and
continuing to match NV1-style entries.
In `@gpu-validation/tasks/nvswitch.yaml`:
- Around line 12-18: Fix the persistent Jinja spacing warning in the
gpu_validation_nvswitch_fm_service_results expression by placing the first
_fm_service_status.results argument on the same line as zip(. Apply this
formatting change in gpu-validation/tasks/nvswitch.yaml lines 12-18,
gpu-validation/tasks/nvswitch_mode0.yaml lines 35-41,
gpu-validation/tasks/nvswitch_mode1.yaml lines 12-18, and
gpu-validation/tasks/nvswitch_mode2.yaml lines 14-20; leave the remaining
expression unchanged.
- Around line 61-68: Update the activated-partition counting expressions in
gpu-validation/tasks/nvswitch.yaml lines 61-68 and
gpu-validation/tasks/nvswitch_mode1.yaml lines 61-68 to use an exact
activated-state match that excludes deactivated entries, while preserving the
existing output and count fact assignments.
---
Duplicate comments:
In `@gpu-validation/tasks/nvswitch_mode1.yaml`:
- Around line 1-84: The mode 1 validation tasks do not inspect fabricmanager.cfg
for the expected FABRIC_MODE. Add a configuration check to nvswitch_mode1.yaml,
matching the established validation used by nvswitch_mode0.yaml, and fail or
report when the configured value does not match the mode 1 expectation. Apply
the corresponding check to nvswitch_mode2.yaml as well if it shares the same
omission.
---
Nitpick comments:
In `@gpu-validation/tasks/nvswitch.yaml`:
- Around line 25-84: The shared collection tasks should be the sole source for
fabric state, fmpm partitions, and NVLink topology facts. Remove the duplicated
collection blocks from nvswitch_mode0.yaml, nvswitch_mode1.yaml, and
nvswitch_mode2.yaml, leaving those files to perform only their mode-specific
checks and FM-service work; preserve the shared task symbols
gpu_validation_nvswitch_fabric_states, gpu_validation_nvswitch_fmpm_output, and
gpu_validation_nvswitch_topo_output.
🪄 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: Organization UI
Review profile: CHILL
Plan: Enterprise
Run ID: 375c5df9-f01a-4b05-9548-ca0f663e2f05
📒 Files selected for processing (18)
gpu-validation/defaults/main.yamlgpu-validation/tasks/nvswitch.yamlgpu-validation/tasks/nvswitch_assertions.yamlgpu-validation/tasks/nvswitch_isolation.yamlgpu-validation/tasks/nvswitch_isolation_assertions.yamlgpu-validation/tasks/nvswitch_mode0.yamlgpu-validation/tasks/nvswitch_mode0_assertions.yamlgpu-validation/tasks/nvswitch_mode0_isolation.yamlgpu-validation/tasks/nvswitch_mode0_isolation_assertions.yamlgpu-validation/tasks/nvswitch_mode1.yamlgpu-validation/tasks/nvswitch_mode1_assertions.yamlgpu-validation/tasks/nvswitch_mode1_isolation.yamlgpu-validation/tasks/nvswitch_mode1_isolation_assertions.yamlgpu-validation/tasks/nvswitch_mode2.yamlgpu-validation/tasks/nvswitch_mode2_assertions.yamlgpu-validation/tasks/nvswitch_mode2_isolation.yamlgpu-validation/tasks/nvswitch_mode2_isolation_assertions.yamlnvswitch_validation.yaml
🚧 Files skipped from review as they are similar to previous changes (12)
- gpu-validation/tasks/nvswitch_mode2_assertions.yaml
- gpu-validation/tasks/nvswitch_mode0_isolation.yaml
- gpu-validation/tasks/nvswitch_mode1_isolation.yaml
- nvswitch_validation.yaml
- gpu-validation/tasks/nvswitch_isolation_assertions.yaml
- gpu-validation/tasks/nvswitch_assertions.yaml
- gpu-validation/tasks/nvswitch_isolation.yaml
- gpu-validation/tasks/nvswitch_mode2_isolation_assertions.yaml
- gpu-validation/tasks/nvswitch_mode1_assertions.yaml
- gpu-validation/defaults/main.yaml
- gpu-validation/tasks/nvswitch_mode0_isolation_assertions.yaml
- gpu-validation/tasks/nvswitch_mode0_assertions.yaml
bogdando
force-pushed
the
nvswitch
branch
3 times, most recently
from
July 13, 2026 13:03
e0e07f0 to
06b32b4
Compare
Provide validations for nwswitch/nvlink for all the supported Fabric Manager modes 0-1-2. Generated-by: claude-4.6-opus-high Signed-off-by: Bohdan Dobrelia <bdobreli@redhat.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Related: RHOSSTRAT-518
Closes: OSPRH-27633
Generated-by: claude-4.6-opus-high]