We have 205 smithi systems used solely as testnodes except a few used as static Jenkins slaves.

smithi{001..128} were purchased and owned by Red Hat. ​(WARRANTY EXPIRED)
smithi{129..205} were purchased by Sage and “loaned” to Red Hat. (WARRANTY VALID THRU 10/20/2019)

Racking Tickets
smithi{001..060} RT 374621
smithi{061..068} RT 390716
smithi{069..128} RT 401820
smithi{129..205} RT 427903

This is untested but claims to have packages available in Fedora 23+ and Ubuntu 16.04 and up.

See https://github.com/linux-nvme/nvme-cli

cd /tmp
git clone https://github.com/linux-nvme/nvme-cli.git
cd nvme-cli/
make
sudo ./nvme smart-log-add /dev/nvme0

Intel provides an RPM (Intel Datacenter Tool) to configure the NVMe cards, update firmware, etc. The firmwares are baked into the RPM so it's important to download the latest zip from Intel whenever possible.

Download the zip file, copy to testnode, unzip and yum localinstall the applicable isdct RPM.

Show NVMe cards

isdct show -intelssd

Update Firmware Example

[root@smithi001 ~]# isdct show -intelssd

- Intel SSD DC P3700 Series CVFT533000EN400BGN -

Bootloader : 8B1B012E
DevicePath : /dev/nvme0n1
DeviceStatus : Healthy
Firmware : 8DV10131
FirmwareUpdateAvailable : Firmware=8DV10171 Bootloader=8B1B0131
Index : 0
ModelNumber : INTEL SSDPEDMD400G4
ProductFamily : Intel SSD DC P3700 Series
SerialNumber : CVFT533000EN400BGN

## NOTE: Use the Index number from above to specify the drive you want to update

[root@smithi001 ~]# isdct load -intelssd 0
WARNING! You have selected to update the drives firmware! 
Proceed with the update? (Y|N): Y
Updating firmware...

- Intel SSD DC P3700 Series CVFT533000EN400BGN -

Status : Firmware Updated Successfully. Please reboot the system.

The NVMe cards have started failing at a faster rate. I'm keeping track of when and how often to see if there's a pattern we can interrupt.

Teuthology will automatically mark a smithi down if it fails to reimage 10 times in a row. This table is used to track if particular machines keep hitting this problem even after BIOS/firmware updates and power resets.

Some partners have donated hardware for us to test. This table shows what and where ended up in smithis.

I (dgalloway) copy ISOs to cobbler.front.sepia.ceph.com:/samba/anonymous.

1. Disable the firewall on cobbler.front.sepia.ceph.com
   1. service iptables stop
2. In the smithi IPMI web UI, under Virtual Media → CD-ROM Image, set the following parameters:
   1. Share Host: 172.21.0.11
   2. Path: \Anonymous\foo.iso
3. Click Save
4. Click Mount
5. Reboot and enter the BIOS, when prompted, by hitting <DEL>
6. In the last BIOS tab, “Save & Exit”, select ATEN Virtual CDROM under Boot Override

As of this writing, the latest BIOS version is X10SRW7.410 (4/10/2017)

1. Mount smithibiosx10srw7410.iso using the instructions above
2. Boot to the Virtual CD
3. Once you get a DOS prompt, run flash.bat X10SRW7.410
4. The first time, it'll put the system in “Flashing Mode” and you'll have to reboot and run this again for the actual BIOS update
5. Shut the system OFF then back on for the BIOS update to complete
6. Supermicro suggests restoring BIOS to default settings then setting back up although I've found some settings get reset automatically anyway (like boot order)