Update dependency prometheus/node_exporter to v1.11.0

[renovate]

This PR contains the following updates:

Package	Update	Change	Pending
prometheus/node_exporter	minor	1.10.2 → 1.11.0	v1.11.1

---

Release Notes

prometheus/node_exporter (prometheus/node_exporter)

v1.11.0: 1.11.0 / 2026-04-04

Compare Source

• [CHANGE] Adds distroless container image. #​3592
• [FEATURE] infiniband: Add hw_counters metrics #​2827
• [FEATURE] kernel_hung: Add collector for kernel hung tasks #​3470
• [FEATURE] mixin: Add "All" option to cluster variable in dashboards #​3509
• [FEATURE] netlink: Add node_network_altnames_info metric #​3541
• [FEATURE] thermal: Add support for CPU temperature metrics on Apple Silicon #​3547
• [FEATURE] bcachefs: Add new collector for bcachefs #​3523
• [ENHANCEMENT] mixin: Make devices in network graphs configurable #​3525
• [ENHANCEMENT] nvme: Add namespace parameters #​3360
• [BUGFIX] netdev: Respect deviceFilter for node_network_address_info metric #​3340
• [BUGFIX] mountstats: Fix multiple transports statistics labeling #​3358
• [BUGFIX] diskstats: Fix panic when using device-include flag only #​3603

---

Configuration

📅 Schedule: (in timezone America/New_York)

• Branch creation
  • "after 2am and before 8am on monday"
• Automerge
  • At any time (no schedule defined)

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR is behind base branch, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about this update again.

---

• If you want to rebase/retry this PR, check this box

---

This PR was generated by Mend Renovate. View the repository job log.

[github-actions]

prometheus/node_exporter (ansible) 1.10.2 -> 1.11.0

Risk: 🔴 Risk

The Deep Dive

Update Scope

Bumps the node_exporter_version Ansible variable from 1.10.2 to 1.11.0 (minor). The geerlingguy.node_exporter role downloads the binary from the GitHub release and installs it to /usr/local/bin/node_exporter on hosts where node_exporter_enabled: true — per ansible/group_vars/pi.yaml and ansible/group_vars/kubernetes.yaml that is pantrypi, garagepi, and k1–k5 (7 hosts). Runtime flags (node_exporter_options) are unchanged: --collector.textfile.directory=/var/lib/node_exporter/textfile_collector --log.level=info.

Features & UX

All new collectors below register as defaultEnabled, so they activate automatically on upgrade unless explicitly disabled via --no-collector.<name>.

• kernel_hung — reads /proc/sys/kernel/hung_task_detect_count, a sysctl first shipped in Linux 6.13. On older kernels the file does not exist; in v1.11.0 this causes the collector to fail loudly on every scrape. See Hazards.
• bcachefs — exposes bcachefs filesystem metrics. Not used on your hosts; no effect.
• infiniband.hw_counters — new InfiniBand hardware counter metrics. Not relevant to your hosts.
• thermal on Apple Silicon — Darwin-only; your hosts run Linux, so no effect.
• node_network_altnames_info — new netlink-derived metric exposing interface altnames. Additive, no config change needed.
• nvme namespace parameters — extra NVMe namespace labels; additive.

No configuration change needed to pick up the additive metrics.

Key Fixes

None of the v1.11.0 bug fixes map to your configuration:

• netdev deviceFilter for node_network_address_info — you do not set --collector.netdev.device-include/exclude.
• mountstats multi-transport labeling — you do not enable the mountstats collector (disabled by default) and no NFS multipath clients are in your inventory.
• diskstats panic with --collector.diskstats.device-include only — requires the device-include flag, which your node_exporter_options does not set.

Newer Versions

v1.11.1 (2026-04-07, 3 days after v1.11.0) ships a single fix: kernel_hung: return ErrNoData when the procfs file is missing. This directly addresses #3612, the regression that v1.11.0 introduces — your hosts are exactly the case the fix targets. The PR body already lists v1.11.1 as Pending (still inside Renovate's 14-day minimumReleaseAge window); Renovate will open a follow-up once it stabilizes. Prefer waiting for that PR rather than merging v1.11.0 now and re-bumping in a week.

Hazards & Risks

Regression: kernel_hung collector errors loudly on all 7 target hosts.

v1.11.0 adds the kernel_hung collector as defaultEnabled. It unconditionally opens /proc/sys/kernel/hung_task_detect_count, a sysctl added to Linux mainline on 2024-10-27 and first released in kernel v6.13 — the file grep in v6.12 finds zero occurrences of the symbol. On kernels without it, every Prometheus scrape logs:

level=ERROR source=collector.go:168 msg="collector failed" name=kernel_hung err="open /proc/sys/kernel/hung_task_detect_count: no such file or directory"

(upstream report 3612). Scrapes still succeed and other metrics are exported, but journald/syslog fills with errors at scrape frequency.

Based on how the target hosts are built in this repo, all 7 run pre-6.13 kernels:

• pantrypi, garagepi — built from scripts/rpi-image-customize with IMAGE_TAG="2025-12-04-raspios-trixie-arm64" (Raspberry Pi OS Trixie, which ships a 6.12.x rpi-kernel).
• k1–k5 — provisioned via PXE preseed (preseed.trixie.cfg / preseed.bookworm.cfg) on Proxmox VMs (ens18 interfaces). Debian Trixie ships Linux 6.12 LTS; Debian Bookworm ships 6.1 LTS. A grep for kernel|linux-image|backport across the os-install role, preseed files, and post-install.sh returned no matches, so no backports kernel is being installed.

Your node_exporter_options does not currently disable it. Workarounds if you proceed with v1.11.0:

• Append --no-collector.kernel_hung to node_exporter_options in ansible/group_vars/all/node_exporter.yaml, or
• Skip this PR and wait for the v1.11.1 Renovate PR, which fixes the collector upstream.

No other breaking changes identified. The new distroless container image does not apply — you install the raw Linux binary, not the container image.

Sources

• node_exporter v1.11.0 release
• node_exporter v1.11.1 release
• Issue #3612: kernel_hung collector fails loudly
• PR #3613: Fix kernel_hung for no data
• PR #3470: Add collector for hung_task_detect_count
• kernel_hung collector source (v1.11.0, defaultEnabled)
• PR #3603: diskstats panic fix
• PR #3592: distroless container image
• Linux commit adding hung_task_detect_count (2024-10-27, first in v6.13)
• scripts/rpi-image-customize (pins raspios trixie arm64 image)
• preseed.trixie.cfg (k-node Debian install)
• preseed.bookworm.cfg (k-node Debian install)

---

🔴 Verdict: Risk

Flagged as renovate:risk: v1.11.0 adds a defaultEnabled kernel_hung collector that errors on every scrape on your kernels, and v1.11.1 (already pending) is a drop-in fix. Close or hold this PR and merge the v1.11.1 Renovate PR once it lands; alternatively, merge now and add --no-collector.kernel_hung to node_exporter_options.

[renovate]

Renovate Ignore Notification

Because you closed this PR without merging, Renovate will ignore this update (1.11.0). You will get a PR once a newer version is released. To ignore this dependency forever, add it to the ignoreDeps array of your Renovate config.

If you accidentally closed this PR, or if you changed your mind: rename this PR to get a fresh replacement PR.