This exporter for the Prometheus monitoring system calls into the ssacli utility to provide metrics for errors reported by HP RAID hardware. Under the hood, it invokes the following command when being scraped:

ssacli ctrl all diag file=/tmp/...

This command generates a ZIP file containing a diagnostic report. The exporter opens the ZIP file and extracts the ADUReport.xml file. It then reports all of the active error conditions from all of the <Errors/> blocks.

There are at least two other Prometheus metrics exporters for HP RAID controllers (1, 2). Both these exporters are built around the idea of parsing the utility's user-readable output using regular expressions, as opposed to relying on the utility's XML output. This makes them a lot more brittle.

Though this exporter only provide metrics for errors for the time being (hpraid_errors), it should be fairly easy to extend it to expose other attributes. All of those metrics should already be part of the ADUReport.xml file that is used by this exporter.

The exporter can be built using Bazel:

bazel build //...

The following command will start the exporter, causing it to listen on TCP port 9423:

./hpraid_exporter

Do note that it expects ssacli to be in $PATH. It also needs to run as a user that is capable of running this command successfully (root).

Example metrics output:

# HELP hpraid_errors Errors in the diagnostic report reported by the RAID controller.
# TYPE hpraid_errors gauge
hpraid_errors{device="ArrayController=Smart Array P410i in Embedded Slot",message="The cache is temporarily disabled.",severity="Warning"} 1
# HELP hpraid_scrape_success Whether scraping the HP RAID controller stats was successful.
# TYPE hpraid_scrape_success gauge
hpraid_scrape_success 1