galexrt
Grafana: changed query field
The temp fields reported by smartctl do not match the grafana query.
Smartctl reports smartmon_temperature_celsius_raw_value
The graph is looking for smartmon_airflow_temperature_cel_value
smartctl version 6.4 2014-10-07 r4002 [x86_64-linux-4.19.0-0.bpo.2-amd64] image digest: 7c54c3be8aa5
Happy to make the PR myself, just not sure if it's a global issue or not
@tdorsey Thanks for raising the issue!
I have gone ahead and updated the Dockerfile to use debian:buster-slim and updated the smartmon.sh script.
Could you please verify that continue to get the smartmon_temperature_celsius_raw_value metrics with the latest image v1.1.0 (quay image is already built, Docker Hub still needs a few more minutes).
A potential fix could be to use the and operator (in my case it seems to work as expected):
avg(smartmon_temperature_celsius_raw_value{instance=~"$instance"} and smartmon_airflow_temperature_cel_value) by (instance, disk)
(https://prometheus.io/docs/prometheus/latest/querying/operators/#logical-set-binary-operators)
Would you be up to open a PR changing the affected queries to use and in them as shown in the example?
Yeah, that's no problem, although I can't guarantee an ETA.
Keeping in mind that I know next to nothing about prometheus vectors, do you know why an or instead of and fails to return data? I'd assume that if smartmon ever returned both airflow* and raw* that the graph would have duplicated data, but using an or operator fails to return either set...
I have looked into the whole thing a bit further, seems that different HDDs might report a different set of SMART values.
In my K8S cluster with the v1.1.0 image (and before that) I'm getting both metrics (smartmon_temperature_celsius_raw_* and smartmon_airflow_temperature_*).
What do you think about adding keeping the Disk Temperature and adding a new panel next to it with the airflow temps?
So like this Smartmon Textfile-1569782184745.txt
(I had to rename the file to end with .txt because GitHub doesn't allow .json uploads)
This isn't a perfect solution, as one graph might be empty depending on the SMART values exposed by the disks, though at least from my perspective it would be a good compromise for the dashboard.
That works for me. I’ll work that into the PR.
This does mean that the chart json would move into this repo though. I don’t know of a way to collaborate on a chart that’s been uploaded to grafana. Is that ok?
@galexrt Sorry for the delay on this. The version of grafana I'm working with has changed json schema formats significantly in the interim, but this version is working for me. Let me know if you see issues
@tdorsey As written in the PR, just update the existing dashboard json file. Please add the Grafana version requirement to the README Grafana section.