Why do the fans ramp up and iDRAC show nothing after I fit a non-Dell NVMe SSD in my PowerEdge?

1. 1.Start in the OS, not in iDRAC. List the NVMe devices the kernel found. This tells you whether the drive and its PCIe link work, independently of iDRAC.

   nvme list

   ReadingA row with the expected Model, SN and FW Rev means the drive is detected and working, and the problem is iDRAC inventory or fan policy. No row at all points at the bay, backplane, cabling or BIOS NVMe mode: go to the third path.
2. 2.Read health from the drive itself, because iDRAC may show none for a non-Dell drive. Use the controller node that nvme list printed. The field names below are those of the libnvme SMART / Health log structure; nvme-cli's printed labels differ slightly from them.

   nvme smart-log /dev/nvme0

   Readingcritical_warning: indicates critical warnings for the state of the controller, so any non-zero value needs attention. temperature: the composite temperature. avail_spare: a normalized 0-100% of remaining spare capacity; when it falls below spare_thresh an asynchronous event may be raised. percent_used: a vendor-specific estimate of life used; 100 means the estimated endurance is consumed but may not indicate a failure. media_errors: counts unrecovered data integrity errors the controller detected.
3. 3.Read the firmware the drive holds. iDRAC will not report it if it does not inventory the drive.

   nvme fw-log /dev/nvme0

   ReadingPrints the Firmware Slot log. The man page names no output fields, so read it alongside the FW Rev column of nvme list. We found no Dell document covering firmware for non-Dell drives; look to the drive maker's own tools.
4. 4.Now ask iDRAC what it inventories. The -p list takes at most 10 property names.

   racadm storage get pdisks -o -p Manufacturer,ProductId,Revision,FailurePredicted,RemainingRatedWriteEndurance,BusProtocol,State

   ReadingIn Dell's examples NVMe drives appear behind PCIeExtender.Slot.N controllers. A drive that nvme list shows but this list omits, or lists with empty fields, is an iDRAC inventory gap for an unsupported drive, not a missing drive.
5. 5.Read the gap against Dell's own wording. Dell's fan-noise article says Dell cannot support a third-party device or guarantee its function within the system, so for a non-Dell drive the OS readings above are your health record. On systems using the PERC S140, Dell's S140 guide supports only Dell-compliant NVMe PCIe SSDs, and in its list of S140 physical-disk properties the Hardware Vendor field 'Displays Unknown option, when a Toshiba NVMe drive is connected'. That note is about the S140's own property view and about Toshiba drives; it says nothing about non-Dell drives as a class.

1. 1.Open the iDRAC web interface, go to the Lifecycle Log and search for message ID PCI3018. We have not verified the racadm syntax for reading this log, so use the web interface for this check.
   ReadingPCI3018 reads 'New PCI card(s) have been detected in the system. Fan speeds may have changed to add additional cooling to the cards.' and means the server is applying its third-party card response. Dell's white paper says that if no such message is logged, the server is not treating the card as third-party and the disable commands will have no effect on fan speed.
2. 2.Dell's documents describe this response for third-party PCIe cards. A drive on a PCIe adapter card sits in a PCIe slot. We found no Dell document saying the response is triggered by a U.2 drive in a front bay, which is why the PCI3018 check comes first. Dell's 13G article says the response targets a maximum of 55C inlet air to the PCIe card region.
3. 3.iDRAC8-era systems: Dell's white paper, written for iDRAC8-era servers, gives this racadm command to read the current state of the third-party PCI fan response. (The 13G model list, R230, R330, R430, R530, R630, R730, R730xd, R830 and R930, comes from Dell's KB 000135682, which covers the ipmitool method below.)

   racadm get system.thermalsettings.ThirdPartyPCIFanResponse

   ReadingReturns the current state. On iDRAC9 systems forum users report this object is rejected with SWC0244 or 'Invalid object specified'; use the iDRAC9 step below instead.
4. 4.iDRAC8-era systems only, and only if PCI3018 is logged. Before you run it: disabling removes the extra cooling Dell provisions for the card, so the drive may run hotter. Dell's white paper says 'Changing default fan response to PCIe cards should be done with appropriate understanding of the PCIe cooling needs of the cards present in the system.' The case it gives for disabling is that 'a card that has its own cooling fan may not need additional cooling'; a fanless M.2 adapter card is not that case. This is a reversible cooling setting, not a data change. In Dell's white paper the value 1 DISABLES the third-party card response (the default is enabled) and 0 re-enables it. Many forum posts use 0 to disable, which is the opposite of Dell's document: use Dell's values.

   racadm set system.thermalsettings.ThirdPartyPCIFanResponse 1

   ReadingRead the setting back with the get command above, then re-run nvme smart-log /dev/nvme0 and watch temperature and warning_temp_time (minutes operated above the warning composite temperature threshold). If the drive runs hot, set the value back to 0. Dell's white paper says fan speed above baseline that other components drive is not manually adjustable, so this does not lower fans for any other reason.
5. 5.13G alternative from Dell's KB 000135682. IPMI over LAN must be enabled in iDRAC, and you need an administrator-level IPMI LAN user. Query the status first.

   ipmitool -I lanplus -H <IPADDRESS> -U <USERNAME> -P <PASSWORD> raw 0x30 0xce 0x01 0x16 0x05 0x00 0x00 0x00

   Reading'16 05 00 00 00 05 00 01 00 00' means the response is Disabled; '16 05 00 00 00 05 00 00 00 00' means Enabled.
6. 6.13G only, and only if PCI3018 is logged. Before you run it: this removes the extra cooling Dell provisions for the card, so the drive may run hotter. Dell's white paper says changing the default fan response 'should be done with appropriate understanding of the PCIe cooling needs of the cards present in the system', and its example of a card that may not need the extra cooling is one with its own cooling fan. Dell's KB says the command only removes the fan response associated with the addition of a third-party PCIe card. It changes cooling, not data, and the enable command in the next step reverses it.

   ipmitool -I lanplus -H <IPADDRESS> -U <USERNAME> -P <PASSWORD> raw 0x30 0xce 0x00 0x16 0x05 0x00 0x00 0x00 0x05 0x00 0x01 0x00 0x00

   ReadingRun the status query again: '16 05 00 00 00 05 00 01 00 00' confirms Disabled. Then re-run nvme smart-log /dev/nvme0 and watch temperature and warning_temp_time; if the drive runs hot, re-enable the response.
7. 7.To reverse it on a 13G server, run the enable command from Dell's KB 000135682.

   ipmitool -I lanplus -H <IPADDRESS> -U <USERNAME> -P <PASSWORD> raw 0x30 0xce 0x00 0x16 0x05 0x00 0x00 0x00 0x05 0x00 0x00 0x00 0x00

   ReadingRun the status query again: '16 05 00 00 00 05 00 00 00 00' confirms Enabled.
8. 8.iDRAC9 (14G and later): the iDRAC9 RACADM guide lists a per-slot System.pcieSlotLFM group as new for iDRAC9, with 3rdPartyCard read-only and LFMMode and CustomLFM read/write. Read it per slot.

   racadm get System.PCIeSlotLFM

   ReadingLook at the read-only 3rdPartyCard object for each slot. We have not read a Dell document giving the values of 3rdPartyCard or LFMMode, so this page prints none; change the setting in the web interface instead.
9. 9.To change it on iDRAC9, go to Configuration > System Settings > Hardware Settings > Cooling Configuration, open the PCIe airflow settings, select Custom mode to enable the custom LFM option, enter the LFM value, and reboot to apply. Dell describes these settings for custom high powered PCIe cards, and they are not available for M.2 drives connected through direct risers or BOSS. No Dell document we read says they apply to a U.2 drive in a front bay.
10. 10.iDRAC9 thermal settings also include a Maximum PCIe Inlet Temperature Limit (default 55°C, with a 45°C option for third-party PCIe cards that need a lower inlet temperature). Dell says the thermal settings feature requires an iDRAC Enterprise or Datacenter licence, so check your licence before looking for the menu.
11. 11.If no PCI3018 is logged and the fans stay high, check the drive's own temperature in nvme smart-log: temperature, and warning_temp_time for minutes operated above the warning composite temperature threshold (libnvme struct names; nvme-cli's printed labels differ slightly). Dell has documented iDRAC raising fan RPM on inaccurate temperature data from an NVMe drive, fixed by a drive firmware update; that article concerns a Dell-branded Micron 7500, not a third-party drive.

    nvme smart-log /dev/nvme0

    ReadingAn implausible temperature points at the drive's own reporting; check the drive maker for firmware. A normal temperature with no PCI3018 leaves Dell's general statement that iDRAC may spin fans up for an unknown device.

1. 1.List the storage controllers iDRAC knows about.

   racadm storage get controllers

   ReadingIn Dell's example NVMe bays sit behind PCIeExtender.Slot.N controllers. If no PCIe extender or NVMe controller appears, that points to a chassis with no NVMe-wired bays, so the drive may be the wrong class for the bay.
2. 2.Check that the bay is NVMe-wired. Dell's R740xd manual gives the system as '32 SAS, SATA, Nearline SAS hard drives or SSDs, and up to 24 NVMe drives', and Dell's S140 guide says its drive count depends on the system backplane configuration. No Dell document we read says which bays are NVMe-capable, so confirm against your backplane and your model's own documentation, and move the drive to an NVMe bay before anything else.
3. 3.Rule out a drive of the wrong class. Dell's S140 troubleshooting for a physical disk that is not visible includes checking whether it is a SAS drive.
4. 4.On systems using the PERC S140, read NVMe Mode in System Setup. Reading it is harmless. Changing it is not: Dell warns that changing NVMe Mode to RAID causes data loss on the NVMe SSD, so back up anything on the drive first. RAID configuration and boot from an NVMe virtual disk are supported only in UEFI boot mode, and Dell's S140 guide supports only Dell-compliant NVMe PCIe SSDs.
   Destructive. This command destroys what is on the device and cannot be undone. Confirm the device path twice and have a backup of anything you still need before running it.

   F2 System Setup > System BIOS > NVMe Settings > NVMe Mode (read the current value; changing Non RAID <-> RAID)

   ReadingNon RAID presents the drives as plain NVMe to the OS. RAID hands them to the S140, which needs its driver and UEFI boot mode.

Dell's US Limited Hardware Warranty covers defects in materials and workmanship in Dell-branded hardware products. Under 'What is not covered by this limited hardware warranty?' it lists 'Commercial hardware products that use, or in which have been installed, products or components that have not been provided by Dell.' It also lists problems that result from 'Using accessories, parts or components not supplied by Dell.' This page does not decide whether your PowerEdge counts as commercial hardware; read the warranty terms that apply to your system.

On systems using the PERC S140, Dell's guide supports only Dell-compliant NVMe PCIe SSDs. We have no Dell document stating that a Dell-branded drive changes the fan behaviour described here, and none naming a specific third-party drive as supported in a PowerEdge model.

Those are iDRAC9 systems. The iDRAC9 RACADM guide lists the per-slot System.pcieSlotLFM group as new for iDRAC9, and we found no iDRAC9 Dell document listing ThirdPartyPCIFanResponse. Forum users report the object returning SWC0244 or 'Invalid object specified' there. Use racadm get System.PCIeSlotLFM and the PCIe airflow settings instead.