[Question] grafana dashboard

[erik78se]

Is there a good grafana dashboard that works well with this exporter?

[georgesouzafarias]

That's a nice idea.
I'm looking for it right now.

[lyz-code]

Did you find any good dashboard?

[erik78se]

We put together a custom one, which I would really like to see someone with more skills than us have put together.

[tarkhil]

So still nothing?

[erik78se]

We still not have found one. Still using the one we cooked up.

[tarkhil]

Can you please add it to github so someone could improve it?

[erik78se]

Can you please add it to github so someone could improve it?

We're in the process of upgrading it some. We'll make some effort for sure to push it here.

[tnyeanderson]

Even if it isn't perfect, I think it's beneficial to at least put it in a draft PR so others can use it and improve it themselves. I'm sure it is "good enough" for most of us, at least as a starting point :)

[wannabe-quant]

any official or unoffical dashboards? someone willing to paste their json files? see the metrics in prom but nothing on dashboard. https://grafana.com/grafana/dashboards/12586-zfs/

Edit:
Found one that works...but its from node_exporter not zfs_exporter. https://grafana.com/grafana/dashboards/7968-node-zfs-stats/

[d-uzlov]

I can share my dashboards for ZFS: https://github.com/d-uzlov/k8s-homelab/tree/master/metrics/zfs-exporter/dashboards
They has some dependencies on my systems, but I think they should mostly work with generic metrics.

• ZFS ARC statistics needs node_exporter
• ZFS vdev needs both pdf/zfs_exporter and siebenmann/zfs_exporter (or equivalent, there are some forks)
• ZFS Pool, ZFS Dataset List and ZFS Dataset Details primarily need node_exporter and pdf/zfs_exporter, but some graphs also use siebenmann/zfs_exporter

P.S. On second thought, I think there is one hard requirement that may not be satisfied with generic metrics: you need to drop port from instance="address:port" label, so that metrics from different sources have the same instance value. I'm not sure how much of the dashboards will break without it.

[d-uzlov]

I can share my dashboards for ZFS

I have since improved a bit on the dashboards.
I'm still not completely happy with them, but I think they are pretty useful in current state.

A few notes:

• I have several recording rules for things that are costly to calculate on the fly (histogram quantiles). Previously I also used recording rules for trim and scrub ETA, but latest version stopped using it.
• I have enabled almost all metrics on the pdf/zfs_exporter. Default set of metrics is very limited
• A lot of the interesting metrics are from siebenmann/zfs_exporter, dashboards will be half-empty without it
• • This includes: vdev, TRIM, scrub/resilver and a few other metrics

My full setup can be found near dashboards, if you are interested. There are systemd service files for exporters, scrape configs, recording rules, a few alerting rules.

Here are screenshots:

pool overview

dataset list

vdev list

individual disks