Add erofs in default excluded fs type

[Nemric]

resolve #3309
erofs should be considered as squashfs in default exporter behavior

[yeazelm]

This looks like a great improvement to avoid reporting on read only filesystems!

[rcosperec]

Any ETA on this ? EKS 1.33 and subsequent bottlerocket upgrades made this a much needed change

[misterek]

This would be quite helpful to get landed. Thanks!

[thomasjcf21]

We are also running into this problem on EKS 1.33 and Bottlerocket. All of our nodes are screaming about running out of disk space :/

[thomasjcf21]

@misterek if you deploy this using kube-prometheus-stack they've fixed this in chart version 73.2.3. You can see what was changed here, and how to do it manually if you aren't using that:

https://github.com/prometheus-community/helm-charts/blob/kube-prometheus-stack-73.2.3/charts/kube-prometheus-stack/values.yaml

[septatrix]

@discordianfish is there anything left before this can get merged? E.g. a second approval by a team member etc.?

[serg-sno]

@SuperQ, @discordianfish, could you please approve this PR?

[Alex-Bantiuc]

Is there any ETA when the fix will be merged? 😞

[Nemric]

@misterek if you deploy this using kube-prometheus-stack they've fixed this in chart version 73.2.3. You can see what was changed here, and how to do it manually if you aren't using that:

https://github.com/prometheus-community/helm-charts/blob/kube-prometheus-stack-73.2.3/charts/kube-prometheus-stack/values.yaml

I did the same PR in Victoria Metrics that I use ;)
VictoriaMetrics/helm-charts#2107
You should be able to override the --collector.filesystem.ignored-fs-types with PR contents

[abuhrovyi]

VictoriaMetrics reacted faster than Prometheus - something odd is going on 😅

[Nemric]

VictoriaMetrics reacted faster than Prometheus - something odd is going on 😅

You can't compare them this way ^^ VM uses the prometheus node-exporter and I did update "cli parameters"
Here I'm updating the "source code" with default parameters that will impact a lot of users in a good and/or bad way

[abuhrovyi]

VictoriaMetrics reacted faster than Prometheus - something odd is going on 😅

You can't compare them this way ^^ VM uses the prometheus node-exporter and I did update "cli parameters" Here I'm updating the "source code" with default parameters that will impact a lot of users in a good and/or bad way

But I also haven't seen a similar change in the Prometheus stack Helm chart, as was done in VM, have I? 😉

[Nemric]

VictoriaMetrics reacted faster than Prometheus - something odd is going on 😅

You can't compare them this way ^^ VM uses the prometheus node-exporter and I did update "cli parameters" Here I'm updating the "source code" with default parameters that will impact a lot of users in a good and/or bad way

But I also haven't seen a similar change in the Prometheus stack Helm chart, as was done in VM, have I? 😉

Yes, I did push a PR to VM helm repo but not in Prometheus helm repo ... feel free to do it, I think PR are always welcome :)