HPE ProLiant DL20 Gen9

HPE · 1U rack server, two 3.5-inch bays · HPE ProLiant DL20 Gen9

what it is

A rack server one unit high: 434 × 43.2 × 383 mm, two 3.5-inch drive bays, one power supply and one expansion card riser.

The chassis comes in two drive layouts — two 3.5-inch drives or four 2.5-inch drives. The enclosure outline is the same, the front layout is completely different. This description covers the two-drive version.

The server has a built-in iLO 4 — a separate management computer that also works when the server's operating system does not boot or the server is powered off. It has its own network address and its own hardware address.

what we read from it

From the iLO 4 controller (firmware 2.73):

• Temperatures from each sensor separately — air inlet, processor, memory, voltage regulators.
• Fans — three, each with status and duty cycle in percent.
• Power supply — status and power, plus the power draw of the whole server.
• Aggregate health status: BIOS, fans, temperatures, power, processor.
• Firmware inventory — BIOS, iLO and other components with versions.
• Logs: the server's hardware events (IML) and events of iLO itself (IEL).

From the operating system installed on the server — if it is monitored — come CPU load, memory and disk usage, uptime and a full set of data about the network link.

port layout

Front panel: two 3.5-inch drive bays, above them an upper-row bay (in this chassis it holds an optical drive), a power button with status LEDs and a USB port.

Rear panel: two RJ-45 network ports, the expansion card riser, USB ports, a video port and the power supply bay.

The management controller shares a port with the first data port — the vendor labels it "NIC1/shared iLO connector". This is the most important thing about this model when planning: cutting the data cable also cuts out-of-band management. In the larger DL380 Gen9 it is the other way round — there iLO has a dedicated port. An assumption carried over from the larger model promises access that will not be there after a failure.

what to watch out for

• iLO can be read from inside the server, via an internal channel. Linux with the vendor's tools, run with administrator rights, returns full hardware health without network traffic and without an iLO account.
• SNMP v1/v2c on iLO 4 has a separate switch that tools run from the server's operating system do not set. The community and the trap receiver are saved, but the agent still does not respond — it is enabled in the iLO web interface or via Redfish.
• Rebooting iLO alone does not affect the server. The controller comes back up after about 90 seconds, the operating system and virtual machines keep running. During that time the iLO web interface does not respond.

what we cannot read

• Which of the two data ports is in use. That depends on the installation, not on the model.
• iLO status when the cable in the shared port is unplugged — see above.

full parameter list

A full parameter list for this model is not here yet — the ones described above are those we have read and verified. A missing list does not mean the device exposes nothing more.