This is a simple iDRAC (and iLO and XClarity) exporter for Prometheus. The exporter uses the Redfish API to communicate with iDRAC and it supports the regular /metrics
endpoint to expose metrics from the host passed via the target
parameter. For example, to scrape metrics for an iDRAC instance on the IP address 123.45.6.78
call the following URL addresse.
http://localhost:9348/metrics?target=123.45.6.78
Every time the exporter is called with a new target, it tries to establish a connection to iDRAC. If the target is unreachable or if the authentication fails, the target will eventually be flagged as invalid, and any subsequent call to that target will simply be ignored and a status code 500 is returned.
In late May 2023 the project was rewritten to use the official Prometheus golang client library. This rewrite also included a number of breaking changes (configuration file format and metrics names). If you are upgrading from an older version, make sure to the read the information in this README again and make the necessary adjustments.
The program supports several different systems, because they all follow the Redfish standard. The exporter has been tested on the following systems.
- HPE iLO 4/5
- Dell iDRAC 9
- Lenovo XClarity
The exporter is written in Go and it can be downloaded and compiled using:
go install github.com/mrlhansen/idrac_exporter/cmd/idrac_exporter@latest
There is a Dockerfile
in the repository for building a container image. To build it locally use:
docker build -t idrac_exporter .
There are also pre-built images available on Docker Hub. To download and run these images, use the following command and replace {tag}
with an actual tag number.
docker run -v /host-path/config.yml:/etc/prometheus/idrac.yml -p 9348:9348 mrlhansen/idrac_exporter:{tag}
Remember to set the listen address to 0.0.0.0
when running inside a container.
In the configuration file for the iDRAC exporter you can specify the bind address and port for the metrics exporter, as well as username and password for all iDRAC hosts. By default the exporter looks for the configuration file in /etc/prometheus/idrac.yml
but the path can be specified using the -config
option.
address: 127.0.0.1 # Listen address
port: 9348 # Listen port
timeout: 10 # HTTP timeout (in seconds) for Redfish API calls
retries: 1 # Number of retries before a target is marked as unreachable
hosts:
123.45.6.78:
username: user
password: pass
default:
username: user
password: pass
metrics:
system: true
sensors: true
power: true
sel: false # iDRAC only
storage: false
memory: false
As shown in the example above, under hosts
you can specify login information for individual hosts via their IP address, otherwise the exporter will attempt to use the login information under default
. Under metrics
you can select what kind of metrics that should be returned, as described in more detail below.
Because the metrics are collected on-demand it can take several minutes to scrape the metrics endpoint, depending on how many metrics groups are selected in the configuration file. For this reason you should carefully select the metrics of interest and make sure Prometheus is configured with a sufficiently high scrape timeout value.
The exporter can expose the metrics listed in the sections below. For all <name>_health
metrics the value has the following mapping.
- 0 = OK
- 1 = Warning
- 2 = Critical
These metrics include power, health, and LED state, total memory size, number of physical processors, BIOS version and machine information.
idrac_system_power_on 1
idrac_system_health{status="OK"} 0
idrac_system_indicator_led_on{state="Lit"} 1
idrac_system_memory_size_bytes 137438953472
idrac_system_cpu_count{model="Intel(R) Xeon(R) Gold 6130 CPU @ 2.10GHz"} 2
idrac_system_bios_info{version="2.3.10"} 1
idrac_system_machine_info{manufacturer="Dell Inc.",model="PowerEdge C6420",serial="abc",sku="xyz"} 1
These metrics include temperature and FAN speeds.
idrac_sensors_temperature{id="0",name="Inlet Temp",units="celsius"} 19
idrac_sensors_fan_speed{id="0",name="FAN1A",units="rpm"} 7912
These metrics include two sets of power readings. The first set is PSU power readings, such as power usage, total power capacity, input voltage and efficiency. Be aware that not all metrics are available on all systems.
idrac_power_supply_output_watts{id="0"} 74.5
idrac_power_supply_input_watts{id="0"} 89
idrac_power_supply_capacity_watts{id="0"} 750
idrac_power_supply_input_voltage{id="0"} 232
idrac_power_supply_efficiency_percent{id="0"} 91
The second set is the power consumption for the entire system (and sometimes also for certain subsystems, such as the CPUs). The first two metrics are instantaneous readings, while the last four metrics are the minimum, maximum and average power consumption as measure over the reported interval.
idrac_power_control_consumed_watts{id="0",name="System Power Control"} 166
idrac_power_control_capacity_watts{id="0",name="System Power Control"} 816
idrac_power_control_min_consumed_watts{id="0",name="System Power Control"} 165
idrac_power_control_max_consumed_watts{id="0",name="System Power Control"} 177
idrac_power_control_avg_consumed_watts{id="0",name="System Power Control"} 166
idrac_power_control_interval_in_minutes{id="0",name="System Power Control"} 1
On iDRAC only, the system event log can also be exported. This is not exactly an ordinary metric, but it is often convenient to be informed about new entries in the event log. The value of this metric is the unix timestamp for when the entry was created (as reported by iDRAC).
idrac_sel_entry{id="1",message="The process of installing an operating system or hypervisor is successfully completed",component="BaseOSBoot/InstallationStatus",severity="OK"} 1631175352
These metrics include information about disk drives in the machine.
idrac_drive_info{id="Disk.Direct.1-1:AHCI.Slot.5-1",manufacturer="MICRON",mediatype="SSD",model="MTFDDAV240TDU",name="SSD 1",protocol="SATA",serial="xyz",slot="1"} 1
idrac_drive_health{id="Disk.Direct.1-1:AHCI.Slot.5-1",status="OK"} 0
idrac_drive_capacity_bytes{id="Disk.Direct.1-1:AHCI.Slot.5-1"} 240057409536
These metrics include information about memory modules in the machine.
idrac_memory_module_info{ecc="MultiBitECC",id="DIMM.Socket.A2",manufacturer="Micron Technology",name="DIMM A2",rank="2",serial="xyz",type="DDR4"} 1
idrac_memory_module_health{id="DIMM.Socket.A2",status="OK"} 0
idrac_memory_module_capacity_bytes{id="DIMM.Socket.A2"} 34359738368
idrac_memory_module_speed_mhz{id="DIMM.Socket.A2"} 2400
For the situation where you have a single idrac_exporter
and multiple iDRACs to query, the following prometheus.yml
snippet can be used.
scrape_configs:
- job_name: idrac
static_configs:
- targets: ['123.45.6.78', '123.45.6.79']
relabel_configs:
- source_labels: [__address__]
target_label: __param_target
- source_labels: [__param_target]
target_label: instance
- target_label: __address__
replacement: exporter:9348
Here 123.45.6.78
and 123.45.6.79
are the iDRACs to query, and exporter:9348
is the address and port where idrac_exporter
is running.