STL-78579

On the RDE NetworkDeviceFunction resource, the SystemGUID and PermanentSystemGUID properties report a synthesized value that does not match the host sys_image_guid and cannot be verified against any host source. This is an informational identity field with no functional impact.

None.

STL-78577

Per-port InfiniBand identity is not discoverable on the RDE Port resource because it exposes only a synthesized Ethernet MAC address and does not report the InfiniBand PortGUID.

None.

STL-78576

Inventory and asset management tools cannot identify the product from Redfish because the Model and SKU properties on the RDE NetworkAdapter and PCIeDevice resources report the raw PCI device ID (0x0001) instead of the product name and SKU.

None.

STL-78574

The link state that RDE reports may be inconsistent or incorrect for management clients because the mapping from InfiniBand physical port states (Offline, Init, Polling) to the Redfish Port LinkStatus enumeration is unverified and undocumented.

None.

STL-78573

RDE management clients may incorrectly identify the fabric as Ethernet. On an InfiniBand HFI, the Redfish/RDE Port resource reports PortProtocol and LinkNetworkTechnology as Ethernet and provides a synthesized MAC address. These values conflict with the sibling NetworkDeviceFunction, which correctly reports InfiniBand.

None.

STL-77756

After updating the SuperNIC firmware using updateAgent, one or more SuperNIC cards in a multi-card system may remain in a "Pending" firmware state following an AC power cycle. Additional AC power cycles do not resolve the issue.

Re-run the updateAgent firmware update on the affected SuperNIC card(s).

To help ensure all cards complete the firmware update on the first attempt, fully power off the system (not just a power cycle) and wait at least one minute before powering back on, allowing all card capacitors to fully discharge.

STL-76776

Network connectivity may fail after hot-swapping cables of different types on the SuperNIC.

Reboot the host to detect the new cables.

STL-76555

HPC service nodes with live user sessions (such as head, NFS, and Lustre nodes) may be disrupted after a SuperNIC firmware update with updateAgent. Activating the new firmware requires a full server power-off; a warm reboot or adapter-level reset will not work.

None.

STL-76323

SuperNICs may intermittently fail to enumerate on the PCIe bus or initialize after a cold boot or power cycle. In affected systems, one or more SuperNICs may be absent from the OS, preventing the node from participating in the fabric until the device is successfully detected and initialized.

Power cycle the server.

STL-75609

The updateAgent tool may become stuck when downgrading SuperNIC firmware.

Use updateAgent -f when downgrading firmware.

STL-75344

PXE boot over IPv4 intermittently fails with PXE-E18: Server response timeout.

None.

STL-74502

When updating SuperNIC firmware with multiple SuperNIC adapters in a server, some cards may fail to properly update.

Disable the port on the failed SuperNIC cards using opaportconfig -p n disable (where “n” is the port number) and perform the update again.

STL-73738

Reloading the ipoib kernel module without also reloading the hfi1 module may cause FM errors.

To avoid this, always reload the hfi1 module along with the ipoib module.

STL-73195

When updating the SuperNIC firmware using updateAgent -v, the following error may occur:

While waiting for a response to PLDM type 05 and command code 14, a message with type 05 and command 15 was received

Retry the updateAgent -v command.

STL-73088

On some dual-port devices, inbound IPoIB traffic may fail after a system reboot, or after restarting the FM and IPoIB interfaces on the FM node.

Ping out from the unreachable interface, and IP traffic will resume.

STL-72988

When updating firmware in a CN5000 SuperNIC in a management node, updateAgent times out when the FM is running at the same time.

Run the updateAgent with the -m option, which will stop the Fabric Manager if it is running, and then automatically start it when firmware update is finished if it had been running.

Refer to the CN5000 Maintenance and Troubleshooting Guide, “Update the SuperNIC Firmware” for the full procedure.

STL-72817

When reading hfi1 port counters from sysfs, you will receive an Invalid Argument return code.

This message can be ignored.

STL-72598

In some cases, when unloading hfi1 or ib_ipoib, the dmesg log may display an error stating:

hfi1_tid_rdma_flush_wait: no rcd for QP

This message can be ignored.

STL-72583

In some cases the dmesg log may contain a message stating:

invalid logical state transition PORT_DOWN > PORT_ACTIVE.

This is cosmetic and will not affect functionality.

STL-71647

On rare occasions, the link may come up at a width less than 4X.

Bounce the link.

STL-77756

After updating the SuperNIC firmware using updateAgent, one or more SuperNIC cards in a multi-card system may remain in a "Pending" firmware state following an AC power cycle. Additional AC power cycles do not resolve the issue.

Re-run the updateAgent firmware update on the affected SuperNIC card(s).

To help ensure all cards complete the firmware update on the first attempt, fully power off the system (not just a power cycle) and wait at least one minute before powering back on, allowing all card capacitors to fully discharge.

STL-76776

Network connectivity may fail after hot-swapping cables of different types on the SuperNIC.

Reboot the host to detect the new cables.

STL-76323

SuperNICs may intermittently fail to enumerate on the PCIe bus or initialize after a cold boot or power cycle. In affected systems, one or more SuperNICs may be absent from the OS, preventing the node from participating in the fabric until the device is successfully detected and initialized.

Power cycle the server.

STL-75609

The updateAgent tool may become stuck when downgrading SuperNIC firmware.

Use updateAgent -f when downgrading firmware.

STL-74502

When updating SuperNIC firmware with multiple SuperNIC adapters in a server, some cards may fail to properly update.

Disable the port on the failed SuperNIC cards using opaportconfig -p n disable (where “n” is the port number) and perform the update again.

STL-73738

Reloading the ipoib kernel module without also reloading the hfi1 module may cause FM errors.

To avoid this, always reload the hfi1 module along with the ipoib module.

STL-73195

When updating the SuperNIC firmware using updateAgent -v, the following error may occur:

While waiting for a response to PLDM type 05 and command code 14, a message with type 05 and command 15 was received

Retry the updateAgent -v command.

STL-73088

On some dual-port devices, inbound IPoIB traffic may fail after a system reboot, or after restarting the FM and IPoIB interfaces on the FM node.

Ping out from the unreachable interface, and IP traffic will resume.

STL-72988

When updating firmware in a CN5000 SuperNIC in a management node, updateAgent times out when the FM is running at the same time.

Run the updateAgent with the -m option, which will stop the Fabric Manager if it is running, and then automatically start it when firmware update is finished if it had been running.

Refer to the CN5000 Maintenance and Troubleshooting Guide, “Update the SuperNIC Firmware” for the full procedure.

STL-72817

When reading hfi1 port counters from sysfs, you will receive an Invalid Argument return code.

This message can be ignored.

STL-72598

In some cases, when unloading hfi1 or ib_ipoib, the dmesg log may display an error stating:

hfi1_tid_rdma_flush_wait: no rcd for QP

This message can be safely ignored.

STL-72583

In some cases the dmesg log may contain a message stating:

invalid logical state transition PORT_DOWN > PORT_ACTIVE.

This is cosmetic and will not affect functionality.

STL-71647

On rare occasions, the link may come up at a width less than 4X.

Bounce the link.

STL-77756

After updating the SuperNIC firmware using updateAgent, one or more SuperNIC cards in a multi-card system may remain in a "Pending" firmware state following an AC power cycle. Additional AC power cycles do not resolve the issue.

Re-run the updateAgent firmware update on the affected SuperNIC card(s).

To help ensure all cards complete the firmware update on the first attempt, fully power off the system (not just a power cycle) and wait at least one minute before powering back on, allowing all card capacitors to fully discharge.

STL-76776

Network connectivity may fail after hot-swapping cables of different types on the SuperNIC.

Reboot the host to detect the new cables.

STL-76323

SuperNICs may intermittently fail to enumerate on the PCIe bus or initialize after a cold boot or power cycle. In affected systems, one or more SuperNICs may be absent from the OS, preventing the node from participating in the fabric until the device is successfully detected and initialized.

Power cycle the server.

STL-75609

The updateAgent tool may become stuck when downgrading SuperNIC firmware.

Use updateAgent -f when downgrading firmware.

STL-74502

When updating SuperNIC firmware with multiple SuperNIC adapters in a server, some cards may fail to properly update.

Disable the port on the failed SuperNIC cards using opaportconfig -p n disable (where “n” is the port number) and perform the update again.

STL-73738

Reloading the ipoib kernel module without also reloading the hfi1 module may cause FM errors.

To avoid this, always reload the hfi1 module along with the ipoib module.

STL-73195

When updating the SuperNIC firmware using updateAgent -v, the following error may occur:

While waiting for a response to PLDM type 05 and command code 14, a message with type 05 and command 15 was received

Retry the updateAgent -v command.

STL-73088

On some dual-port devices, inbound IPoIB traffic may fail after a system reboot, or after restarting the FM and IPoIB interfaces on the FM node.

Ping out from the unreachable interface, and IP traffic will resume.

STL-72988

When updating firmware in a CN5000 SuperNIC in a management node, updateAgent times out when the FM is running at the same time.

Run the updateAgent with the -m option, which will stop the Fabric Manager if it is running, and then automatically start it when firmware update is finished if it had been running.

Refer to the CN5000 Maintenance and Troubleshooting Guide, “Update the SuperNIC Firmware” for the full procedure.

STL-72817

When reading hfi1 port counters from sysfs, you will receive an Invalid Argument return code.

This message can be ignored.

STL-72598

In some cases, when unloading hfi1 or ib_ipoib, the dmesg log may display an error stating:

hfi1_tid_rdma_flush_wait: no rcd for QP

This message can be safely ignored.

STL-72583

In some cases the dmesg log may contain a message stating:

invalid logical state transition PORT_DOWN > PORT_ACTIVE.

This is cosmetic and will not affect functionality.

STL-71647

On rare occasions, the link may come up at a width less than 4X.

Bounce the link.

STL-75192

In some rare cases, the host system may hang on a warm reboot.

12.1.1

STL-74377

In rare cases, nodes may drop from fabric with unreachable CPORT when attempting to delete or reset an IPoIB interface.

12.1.0

STL-73920

System Image GUID is the same on all SuperNICs.

12.1.0

STL-73153

The SuperNIC firmware version is currently displayed differently in different places, such as the system log versus the updateAgent.

12.1.0

STL-72796

iperf2 may occasionally hang with an SDMA status error.

12.1.0

STL-72644

While using a QSFP cable, you may see warnings in host logs regarding RxDmaDataFifoRdCorErr (QSFP-DD lanes not used).

12.1.0

STL-72640

In some cases the dmesg log may contain a message stating:

hfi1_tid_rdma_flush_wait: no rcd for QP

12.1.0

STL-72360

After IMB-RMA Uniband operation, occasional tasks may freeze.

12.1.0

STL-76776

Network connectivity may fail after hot swapping cables of different types on the SuperNIC.

Reboot the host to detect the new cables.

STL-76323

SuperNICs may intermittently fail to enumerate on the PCIe bus or initialize after a cold boot or power cycle. In affected systems, one or more SuperNICs may be absent from the OS, preventing the node from participating in the fabric until the device is successfully detected and initialized.

Power cycle the server.

STL-75884

PXE boot fails after changing the default Fabric Manager SubnetPrefix.

Restore the Fabric Manager SubnetPrefix to the default value 0xfe80000000000000.

STL-75609

The updateAgent tool may become stuck when downgrading SuperNIC firmware.

Use updateAgent -f when downgrading firmware.

STL-75486

In rare cases, the driver may encounter an error as it loads. You will see errors in the dmesg log saying sysfs: cannot create duplicate filename.

Reboot the host and then load the hfi1 driver.

STL-74502

When updating SuperNIC firmware with multiple SuperNIC adapters in a server, some cards may fail to properly update.

Disable the port on the failed SuperNIC cards using opaportconfig -p n disable (where “n” is the port number) and perform the update again.

STL-73738

Reloading the ipoib kernel module without also reloading the hfi1 module may cause FM errors.

To avoid this, always reload the hfi1 module along with the ipoib module.

STL-73709

The opasaquery -o trace -P (GUIDs) and opasaquery -o trace -G (GIDs) commands fail when targeting CN5000 SuperNIC cards.

No workaround.

STL-73440

Running the updateAgent while the Fabric Manager is running may result in a segfault.

Disable the Fabric Manager prior to running the updateAgent.

STL-73195

When updating the SuperNIC firmware using updateAgent -v, the following error may occur:

While waiting for a response to PLDM type 05 and command code 14, a message with type 05 and command 15 was received

Retry the updateAgent -v command.

STL-73153

The SuperNIC firmware version is currently displayed differently in different places, such as the system log versus the updateAgent.

Use updateAgent -V when checking the SuperNIC firmware version.

STL-73088

On some dual-port devices, inbound IPoIB traffic may fail after a system reboot, or after restarting the FM and IPoIB interfaces on the FM node.

Ping out from the unreachable interface, and IP traffic will resume.

STL-72988

STL-72556

When updating firmware in a CN5000 SuperNIC in a management node, updateAgent times out when the FM is running at the same time.

Run the updateAgent with the -m option, which will stop the Fabric Manager if it is running, and then automatically start it when firmware update is finished if it had been running.

Refer to the CN5000 Maintenance and Troubleshooting Guide, “Update the SuperNIC Firmware” for the full procedure.

STL-72817

When reading hfi1 port counters from sysfs, you will receive an Invalid Argument return code.

This message can be ignored.

STL-72598

In some cases, when unloading hfi1 or ib_ipoib, the dmesg log may display an error stating:

hfi1_tid_rdma_flush_wait: no rcd for QP

This message can be safely ignored.

STL-72583

In some cases the dmesg log may contain a message stating:

invalid logical state transition PORT_DOWN > PORT_ACTIVE.

This is cosmetic and will not affect functionality.

STL-71647

On rare occasions, the link may come up at a width less than 4X.

Bounce the link.

STL-74377

In rare cases, nodes may drop from fabric with unreachable CPORT when attempting to delete or reset an IPoIB interface.

Power cycle the affected node.

STL-73920

System Image GUID is the same on all SuperNICs.

No workaround.

STL-73738

Reloading the ipoib kernel module without also reloading the hfi1 module may cause FM errors.

To avoid this, always reload the hfi1 module along with the ipoib module.

STL-73709

The opasaquery -o trace -P (GUIDs) and opasaquery -o trace -G (GIDs) commands fail when targeting CN5000 SuperNIC cards.

No workaround.

STL-73440

Running the updateAgent while the Fabric Manager is running may result in a segfault.

Disable the Fabric Manager prior to running the updateAgent.

STL-73195

When updating the SuperNIC firmware using updateAgent -v, the following error may occur:

While waiting for a response to PLDM type 05 and command code 14, a message with type 05 and command 15 was received

Retry the updateAgent -v command.

STL-73153

The SuperNIC firmware version is currently displayed differently in different places, such as the system log versus the updateAgent.

Use updateAgent -V when checking the SuperNIC firmware version.

STL-73088

On some dual-port devices, inbound IPoIB traffic may fail after a system reboot, or after restarting the FM and IPoIB interfaces on the FM node.

Ping out from the unreachable interface, and IP traffic will resume.

STL-73075

lspci does not enumerate the SuperNIC with ASPM enabled, causing the BIOS screen on Intel D50DNP platforms to lock up.

Disable ASPM in the BIOS.

STL-72988

STL-72556

When updating firmware in a CN5000 SuperNIC in a management node, updateAgent times out when the FM is running at the same time.

Run the updateAgent with the -m option, which will stop the Fabric Manager if it is running, and then automatically start it when firmware update is finished if it had been running.

Refer to the CN5000 Maintenance and Troubleshooting Guide, “Update the SuperNIC Firmware” for the full procedure.

STL-72817

When reading hfi1 port counters from sysfs, you will receive an Invalid Argument return code.

This message can be ignored.

STL-72796

iperf2 may occasionally hang with an SDMA status error.

Set CONNECTED_MODE=yes in ifcfg.

See the CN5000 Performance Tuning Guide for more details.

STL-72644

While using a QSFP cable, you may see warnings in host logs regarding RxDmaDataFifoRdCorErr (QSFP-DD lanes not used).

These messages can be ignored.

STL-72640

STL-72583

In some cases the dmesg log may contain a message stating:

invalid logical state transition PORT_DOWN > PORT_ACTIVE.

This is cosmetic and will not affect functionality.

STL-72598

In some cases, when unloading hfi1 or ib_ipoib, the dmesg log may display an error stating:

hfi1_tid_rdma_flush_wait: no rcd for QP

This message can be safely ignored.

STL-72360

After IMB-RMA Uniband operation, occasional tasks may freeze.

Reset the SuperNIC.

STL-71647

On rare occasions, the link may come up at a width less than 4X.

Bounce the link.