TruenNAS Core can export performance
metrics to a remote Graphite server. However, if you use Prometheus for system
monitoring and you want to ingest the TrueNAS data you need to run an instance
of graphite_exporter to
ingest the data. The mapping file provided here, maps the Graphite metrics
paths to Prometheus names and labels. Specify the path to the file with the
--graphite.mapping-config=truenas-mapping.yml argument when starting
graphite_exporter.
Altough graphite_exporter warns about possible performance issues due to
backtracking in the match expressions, this mapping allows to retrieve labeled
data. For example, truenas_aggregation_cpu{job="truenas", cpu=~"cpu-average"}
will retrieve idle, interrupt, nice, system and user metrics in one query.
Network activity can be retrieved by
rate(truenas_interface_if_octets{job="truenas", interface=~"ix0|bridge0", action="rx"}[$__rate_interval])
I took my inspiration from in this older Forum Post, but I found it a bit diffcult to navigate through the data as labels are hardly used.
Check the Graphite Separate Instances option in System -> Reporting.
By default TrueNAS Core sends the metrics to port 2003.