v0.2.0

Improvements found by reading a fleet of PowerEdge servers live: R440, R740xd2 and DSS8440 (14G), R750, R7525 and R7515 (15G), on iDRAC firmware 7.00 and 7.20, both standalone and OpenManage-managed.

Upgrading: some devices are renamed

Two changes rename devices on the first poll after updating, for the hardware concerned. Any dzVents script that looks those devices up by name needs updating. Devices you renamed by hand are never touched.

• PCIe SSD in Slot 23 in Bay 2 becomes NVMe in Slot 23 in Bay 2
• NonRAID Solid State Disk 0:1:0 becomes NonRAID SSD 0:1:0

Added

Device name prefix and suffix. If you monitor more than one server, both installs create a device called System Health, and a dzVents lookup by name cannot tell them apart. Across the six machines surveyed, 59 of 186 distinct device names collided. The text is used exactly as you type it, so it carries its own separator, and it can contain {servicetag}, {hostname}, {fqdn}, {idrac} or {model}, which the plugin fills in from the server itself.

Duplicate name warnings. Checked once per start, before any device is created, against every device in your Domoticz install. It names the hardware entry that already owns the name. It only warns, and never renames anything automatically to dodge a collision.

GPU temperature without a telemetry licence. GPU metrics are licence-gated, but a card's temperature is usually also in the ordinary sensor list. Where telemetry reports no cards, the plugin now reads any sensor the server tags with the Redfish physical context GPU. It found a card running at 74 °C on an R750 whose telemetry reported no GPU at all. A fallback, never an addition, so no card gets two temperature devices.

Power supply efficiency, from the AC a supply draws against the DC it delivers. A loaded R750 ran at 93.3%; an R440 idling at 32 W managed only 75.0%. No device appears when the figure would be meaningless, so a supply on standby or under a trickle load produces nothing rather than a misleading zero.

Changed

Power Redundancy tells "switched off" apart from "not reported". An empty redundancy group means either a pulled supply or a deliberately non-redundant configuration, and it used to read Not reported either way. It now reads Not redundant (configured) when the server's own policy says so.

Fixed

Wall-socket power was silently lost on servers advertising many telemetry reports. Reports were read in the order the server returned them, up to a fixed budget. A PowerEdge R440 lists 39 and puts PowerMetrics, the only one carrying SystemInputPower, at position 23, so it was never read: Server Power quietly fell back to the mainboard sensor, which excludes the power supplies' own conversion loss. Reports whose name suggests power are now read first, while which report is actually used is still decided by the metrics it contains.

Also

GPU support has now run against real GPU hardware for the first time, on a 7-card DSS8440 and a 4-card R7525. The documentation gains real device-card images for the states that are hard to describe, including a raised fault, a failed supply and a drive its own SMART has flagged.

Full changelog: https://github.com/Rouzax/dell-idrac-domoticz-plugin/blob/main/CHANGELOG.md

v0.1.0

First release.

Monitors a Dell PowerEdge server from Domoticz over the iDRAC's Redfish API, creating devices for the hardware your server actually reports rather than from a fixed list. Read-only by default; power control and the identify LED stay off until you enable them.

Documentation: https://rouzax.github.io/dell-idrac-domoticz-plugin/

Install

cd /opt/domoticz/plugins
git clone -b dist --single-branch https://github.com/Rouzax/dell-idrac-domoticz-plugin.git dell-idrac

Then restart Domoticz and add hardware of type Dell iDRAC Monitor. Without git, download dell-idrac-v0.1.0.zip below and unzip it into your plugins directory; it already contains the dell-idrac folder.

Requires Domoticz with the Python plugin system and Python 3.11 or newer. No third-party runtime dependencies.

What you get

Temperatures, fan speeds, system and per-supply power with an integrated energy counter, CPU / memory / I/O / system utilization, uptime, boot status, chassis intrusion, per-drive and per-volume health, and per-NIC link state. Plus a single System Health tile that shows the iDRAC's own fault text rather than just a colour, and a Power Redundancy device that catches the case where the group is Critical while every individual supply still reports OK.

Where the iDRAC licence allows it, system power also breaks down by subsystem (CPU, memory, storage, fan, PCIe, FPGA) and per GPU. Either an iDRAC Datacenter licence or an OpenManage Enterprise Advanced licence unlocks it.

Fan, temperature and drive-life cards carry a coloured bar built from the server's own thresholds. Those need a Domoticz build including domoticz/domoticz#6968; on an older build everything else works and the bars simply do not appear.

See the changelog for the full list, including the known limitations.