Merge branch 'development' into alex_readme_nodestatus

Author: alexandraBara

docs/PLUGIN_DOC.md

@@ -4,7 +4,7 @@
 
 | Plugin | Collection | Analyzer Args | Collection Args | DataModel | Collector | Analyzer |
 | --- | --- | --- | --- | --- | --- | --- |
-| AmdSmiPlugin | bad-pages<br>firmware --json<br>list --json<br>metric -g all<br>partition --json<br>process --json<br>ras --cper --folder={folder}<br>ras --afid --cper-file {cper_file}<br>static -g all --json<br>static -g {gpu_id} --json<br>topology<br>version --json<br>xgmi -l<br>xgmi -m | **Analyzer Args:**<br>- `check_static_data`: bool — If True, run static data checks (e.g. driver version, partition mode).<br>- `expected_gpu_processes`: Optional[int] — Expected number of GPU processes.<br>- `expected_max_power`: Optional[int] — Expected maximum power value (e.g. watts).<br>- `expected_driver_version`: Optional[str] — Expected AMD driver version string.<br>- `expected_memory_partition_mode`: Optional[str] — Expected memory partition mode (e.g. sp3, dp).<br>- `expected_compute_partition_mode`: Optional[str] — Expected compute partition mode.<br>- `expected_pldm_version`: Optional[str] — Expected PLDM version string.<br>- `l0_to_recovery_count_error_threshold`: Optional[int] — L0-to-recovery count above which an error is raised.<br>- `l0_to_recovery_count_warning_threshold`: Optional[int] — L0-to-recovery count above which a warning is raised.<br>- `vendorid_ep`: Optional[str] — Expected endpoint vendor ID (e.g. for PCIe).<br>- `vendorid_ep_vf`: Optional[str] — Expected endpoint VF vendor ID.<br>- `devid_ep`: Optional[str] — Expected endpoint device ID.<br>- `devid_ep_vf`: Optional[str] — Expected endpoint VF device ID.<br>- `sku_name`: Optional[str] — Expected SKU name string for GPU.<br>- `expected_xgmi_speed`: Optional[list[float]] — Expected xGMI speed value(s) (e.g. link rate).<br>- `analysis_range_start`: Optional[datetime.datetime] — Start of time range for time-windowed analysis.<br>- `analysis_range_end`: Optional[datetime.datetime] — End of time range for time-windowed analysis. | **Collection Args:**<br>- `cper_file_path`: Optional[str] — Path to CPER folder or file for RAS AFID collection (ras --afid --cper-file). | [AmdSmiDataModel](#AmdSmiDataModel-Model) | [AmdSmiCollector](#Collector-Class-AmdSmiCollector) | [AmdSmiAnalyzer](#Data-Analyzer-Class-AmdSmiAnalyzer) |
+| AmdSmiPlugin | bad-pages<br>firmware --json<br>list --json<br>metric -g all<br>partition --json<br>process --json<br>ras --cper --folder={folder}<br>ras --afid --cper-file {cper_file}<br>static -g all --json<br>static -g {gpu_id} --json<br>topology<br>version --json<br>xgmi -l<br>xgmi -m | **Analyzer Args:**<br>- `check_static_data`: bool — If True, run static data checks (e.g. driver version, partition mode).<br>- `expected_gpu_processes`: Optional[int] — Expected number of GPU processes.<br>- `expected_max_power`: Optional[int] — Expected maximum power value (e.g. watts).<br>- `expected_driver_version`: Optional[str] — Expected AMD driver version string.<br>- `expected_memory_partition_mode`: Optional[str] — Expected memory partition mode (e.g. sp3, dp).<br>- `expected_compute_partition_mode`: Optional[str] — Expected compute partition mode.<br>- `expected_firmware_versions`: Optional[dict[str, str]] — Expected firmware versions keyed by amd-smi fw_id (e.g. PLDM_BUNDLE).<br>- `l0_to_recovery_count_error_threshold`: Optional[int] — L0-to-recovery count above which an error is raised.<br>- `l0_to_recovery_count_warning_threshold`: Optional[int] — L0-to-recovery count above which a warning is raised.<br>- `vendorid_ep`: Optional[str] — Expected endpoint vendor ID (e.g. for PCIe).<br>- `vendorid_ep_vf`: Optional[str] — Expected endpoint VF vendor ID.<br>- `devid_ep`: Optional[str] — Expected endpoint device ID.<br>- `devid_ep_vf`: Optional[str] — Expected endpoint VF device ID.<br>- `sku_name`: Optional[str] — Expected SKU name string for GPU.<br>- `expected_xgmi_speed`: Optional[list[float]] — Expected xGMI speed value(s) (e.g. link rate).<br>- `analysis_range_start`: Optional[datetime.datetime] — Start of time range for time-windowed analysis.<br>- `analysis_range_end`: Optional[datetime.datetime] — End of time range for time-windowed analysis. | **Collection Args:**<br>- `cper_file_path`: Optional[str] — Path to CPER folder or file for RAS AFID collection (ras --afid --cper-file). | [AmdSmiDataModel](#AmdSmiDataModel-Model) | [AmdSmiCollector](#Collector-Class-AmdSmiCollector) | [AmdSmiAnalyzer](#Data-Analyzer-Class-AmdSmiAnalyzer) |
 | BiosPlugin | sh -c 'cat /sys/devices/virtual/dmi/id/bios_version'<br>wmic bios get SMBIOSBIOSVersion /Value | **Analyzer Args:**<br>- `exp_bios_version`: list[str] — Expected BIOS version(s) to match against collected value (str or list).<br>- `regex_match`: bool — If True, match exp_bios_version as regex; otherwise exact match. | - | [BiosDataModel](#BiosDataModel-Model) | [BiosCollector](#Collector-Class-BiosCollector) | [BiosAnalyzer](#Data-Analyzer-Class-BiosAnalyzer) |
 | CmdlinePlugin | cat /proc/cmdline | **Analyzer Args:**<br>- `required_cmdline`: Union[str, List] — Command-line parameters that must be present (e.g. 'pci=bfsort').<br>- `banned_cmdline`: Union[str, List] — Command-line parameters that must not be present.<br>- `os_overrides`: Dict[str, nodescraper.plugins.inband.cmdline.cmdlineconfig.OverrideConfig] — Per-OS overrides for required_cmdline and banned_cmdline (keyed by OS identifier).<br>- `platform_overrides`: Dict[str, nodescraper.plugins.inband.cmdline.cmdlineconfig.OverrideConfig] — Per-platform overrides for required_cmdline and banned_cmdline (keyed by platform). | - | [CmdlineDataModel](#CmdlineDataModel-Model) | [CmdlineCollector](#Collector-Class-CmdlineCollector) | [CmdlineAnalyzer](#Data-Analyzer-Class-CmdlineAnalyzer) |
 | DeviceEnumerationPlugin | powershell -Command "(Get-WmiObject -Class Win32_Processor &#124; Measure-Object).Count"<br>lspci -d {vendorid_ep}: &#124; grep -i 'VGA\&#124;Display\&#124;3D' &#124; wc -l<br>powershell -Command "(wmic path win32_VideoController get name &#124; findstr AMD &#124; Measure-Object).Count"<br>lscpu<br>lshw<br>lspci -d {vendorid_ep}: &#124; grep -i 'Virtual Function' &#124; wc -l<br>powershell -Command "(Get-VMHostPartitionableGpu &#124; Measure-Object).Count" | **Analyzer Args:**<br>- `cpu_count`: Optional[list[int]] — Expected CPU count(s); pass as int or list of ints. Analysis passes if actual is in list.<br>- `gpu_count`: Optional[list[int]] — Expected GPU count(s); pass as int or list of ints. Analysis passes if actual is in list.<br>- `vf_count`: Optional[list[int]] — Expected virtual function count(s); pass as int or list of ints. Analysis passes if actual is in list. | - | [DeviceEnumerationDataModel](#DeviceEnumerationDataModel-Model) | [DeviceEnumerationCollector](#Collector-Class-DeviceEnumerationCollector) | [DeviceEnumerationAnalyzer](#Data-Analyzer-Class-DeviceEnumerationAnalyzer) |
@@ -1843,7 +1843,7 @@ Check sysctl matches expected sysctl details
 - **expected_driver_version**: `Optional[str]` — Expected AMD driver version string.
 - **expected_memory_partition_mode**: `Optional[str]` — Expected memory partition mode (e.g. sp3, dp).
 - **expected_compute_partition_mode**: `Optional[str]` — Expected compute partition mode.
-- **expected_pldm_version**: `Optional[str]` — Expected PLDM version string.
+- **expected_firmware_versions**: `Optional[dict[str, str]]` — Expected firmware versions keyed by amd-smi fw_id (e.g. PLDM_BUNDLE).
 - **l0_to_recovery_count_error_threshold**: `Optional[int]` — L0-to-recovery count above which an error is raised.
 - **l0_to_recovery_count_warning_threshold**: `Optional[int]` — L0-to-recovery count above which a warning is raised.
 - **vendorid_ep**: `Optional[str]` — Expected endpoint vendor ID (e.g. for PCIe).

nodescraper/plugins/inband/amdsmi/amdsmi_analyzer.py

@@ -534,18 +534,14 @@ def _format_static_mismatch_payload(
             "per_gpu": per_gpu_list,
         }
 
-    def check_pldm_version(
+    def check_firmware_versions(
         self,
         amdsmi_fw_data: Optional[list[Fw]],
-        expected_pldm_version: Optional[str],
-    ):
-        """Check expected pldm version
-
-        Args:
-            amdsmi_fw_data (Optional[list[Fw]]): data model
-            expected_pldm_version (Optional[str]): expected pldm version
-        """
-        PLDM_STRING = "PLDM_BUNDLE"
+        expected_firmware_versions: dict[str, str],
+    ) -> None:
+        """Check that each GPU reports the expected version for each ``fw_id``."""
+        if not expected_firmware_versions:
+            return
         if amdsmi_fw_data is None or len(amdsmi_fw_data) == 0:
             self._log_event(
                 category=EventCategory.PLATFORM,
@@ -554,30 +550,37 @@ def check_pldm_version(
                 data={"amdsmi_fw_data": amdsmi_fw_data},
             )
             return
-        mismatched_gpus: list[int] = []
-        pldm_missing_gpus: list[int] = []
+        mismatches: list[dict[str, object]] = []
+        missing: list[dict[str, object]] = []
         for fw_data in amdsmi_fw_data:
             gpu = fw_data.gpu
             if isinstance(fw_data.fw_list, str):
-                pldm_missing_gpus.append(gpu)
+                for fw_id in expected_firmware_versions:
+                    missing.append({"gpu": gpu, "fw_id": fw_id})
                 continue
-            for fw_info in fw_data.fw_list:
-                if PLDM_STRING == fw_info.fw_id and expected_pldm_version != fw_info.fw_version:
-                    mismatched_gpus.append(gpu)
-                if PLDM_STRING == fw_info.fw_id:
-                    break
-            else:
-                pldm_missing_gpus.append(gpu)
+            actual_by_id = {item.fw_id: item.fw_version for item in fw_data.fw_list}
+            for fw_id, expected_ver in expected_firmware_versions.items():
+                if fw_id not in actual_by_id:
+                    missing.append({"gpu": gpu, "fw_id": fw_id})
+                elif actual_by_id[fw_id] != expected_ver:
+                    mismatches.append(
+                        {
+                            "gpu": gpu,
+                            "fw_id": fw_id,
+                            "expected": expected_ver,
+                            "actual": actual_by_id[fw_id],
+                        }
+                    )
 
-        if mismatched_gpus or pldm_missing_gpus:
+        if mismatches or missing:
             self._log_event(
                 category=EventCategory.FW,
-                description="PLDM Version Mismatch",
+                description="Firmware version mismatch",
                 priority=EventPriority.ERROR,
                 data={
-                    "mismatched_gpus": mismatched_gpus,
-                    "pldm_missing_gpus": pldm_missing_gpus,
-                    "expected_pldm_version": expected_pldm_version,
+                    "expected_firmware_versions": expected_firmware_versions,
+                    "mismatches": mismatches,
+                    "missing": missing,
                 },
             )
 
@@ -661,8 +664,9 @@ def check_expected_xgmi_link_speed(
         if expected_xgmi_speed is None or len(expected_xgmi_speed) == 0:
             self._log_event(
                 category=EventCategory.IO,
-                description="Expected XGMI speed not configured, skipping XGMI link speed check",
-                priority=EventPriority.WARNING,
+                description=("Expected XGMI link speed not set; skipping XGMI link speed analysis"),
+                priority=EventPriority.INFO,
+                console_log=True,
             )
             return
 
@@ -778,8 +782,8 @@ def analyze_data(
                 args.expected_compute_partition_mode,
             )
 
-        if args.expected_pldm_version:
-            self.check_pldm_version(data.firmware, args.expected_pldm_version)
+        if args.expected_firmware_versions:
+            self.check_firmware_versions(data.firmware, args.expected_firmware_versions)
 
         if data.cper_data:
             self.analyzer_cpers(

nodescraper/plugins/inband/amdsmi/amdsmi_collector.py

@@ -475,7 +475,8 @@ def _get_amdsmi_data(
             return None
 
         try:
-            return AmdSmiDataModel(
+            fw_ids = args.analysis_firmware_ids if args and args.analysis_firmware_ids else None
+            base = AmdSmiDataModel(
                 version=version,
                 gpu_list=gpu_list,
                 process=processes,
@@ -489,7 +490,10 @@ def _get_amdsmi_data(
                 xgmi_link=xgmi_link or [],
                 cper_data=cper_data,
                 cper_afids=cper_afids,
+                analysis_firmware_ids=fw_ids,
+                analysis_ref=None,
             )
+            return base.model_copy(update={"analysis_ref": base.build_analysis_ref()})
         except ValidationError as err:
             self.logger.warning("Validation err: %s", err)
             self._log_event(
@@ -763,7 +767,9 @@ def get_firmware(self) -> Optional[list[Fw]]:
             normalized: list[FwListItem] = []
             for e in fw_list_raw:
                 if isinstance(e, dict):
-                    fid = e.get("fw_name")
+                    fid = e.get("fw_id")
+                    if fid is None:
+                        fid = e.get("fw_name")
                     ver = e.get("fw_version")
                     normalized.append(
                         FwListItem(