Skip to content

node_filesystem_readonly seems to be reporting _everything_ as read-write #3576

@io7m

Description

@io7m

Hello!

Possibly related to #3484 although the symptom in my case seems a little more extreme. The node_filesystem_readonly always reports 0 indicating that all filesystems are read-write.

In my case, most read-only filesystems are archival ZFS filesystems set as read-only. For example:

$ grep vanilla /proc/1/mounts
storage/backup/archived/vanilla /backup/archived/vanilla zfs ro,seclabel,relatime,xattr,noacl,casesensitive 0 0
$ mount | grep vanilla
storage/backup/archived/vanilla on /backup/archived/vanilla type zfs (ro,relatime,seclabel,xattr,noacl,casesensitive)
node_filesystem_readonly{device="storage/backup/archived/vanilla"}  => 0

This has caused a lot of alerting to start complaining, because it thinks those filesystems are full (because they're resized with a quota down to 100% usage before being made read-only, and read-only filesystems are normally excluded from alerting here).

I'm on:

$ /usr/bin/prometheus-node-exporter --version
node_exporter, version 1.10.2 (branch: tarball, revision: 3.el9)
  build user:       
  build date:       20260130
  go version:       go1.25.5 (Red Hat 1.25.5-2.el9_7)
  platform:         linux/amd64
  tags:             rpm_crashtraceback,libtrust_openssl,netgo,osusergo,static_build
$ uname -a
Linux backup02 5.14.0-611.36.1.el9_7.x86_64 #1 SMP PREEMPT_DYNAMIC Tue Mar 3 17:30:12 UTC 2026 x86_64 x86_64 x86_64 GNU/Linux

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions