Stakkr Observer
Cockpit-based observability for the Stakkr host resource model, packaged as a standalone plugin and published here as the UI landing page.
The observer is the supported Stakkr UI path. It answers the live-state question that the workflow docs deliberately do not:
note
Is the current host resource policy behaving the way Stakkr intends right now?
What It Shows
- current host policy state
- CPU performance domains and weights
- CPU pool topology and per-CPU utilization
- host memory overview, including THP, KSM, and zram state
- memory management overhead from
ksmd,kswapd*, andkcompactd*
Where It Fits
Use the observer after the host foundation or cluster workflows are already in motion. It is not a replacement for the operator entrypoints.
- For the default host CPU model, start with Shared Execution Pool
- For the alternate experiment, use Clock Frequency Tiering
- For repo-wide navigation, go back to the Documentation Map
Install Paths
From the built RPM at the repo root:
cd /path/to/stakkr
sudo dnf install -y ./cockpit/stakkr-observer/rpmbuild/RPMS/noarch/cockpit-stakkr-observer-1.0.0-1.el10.noarch.rpm
From source into Cockpit's plugin directory:
sudo mkdir -p /usr/share/cockpit/stakkr-observer
sudo rsync -av /path/to/stakkr/cockpit/stakkr-observer/ /usr/share/cockpit/stakkr-observer/
Panels
| Panel | Purpose |
|---|---|
| Current State | Live host policy state, host reserve, emulator placement, and guest domain placement |
| CPU Performance Domains | Scope weights, allowed CPUs, vCPU pinning, and emulator pinning |
| CPU Pool Topology | Per-CPU role map, utilization, and current frequency |
| Memory Overview | Host memory totals, THP state, KSM savings, zram state, and swap usage |
| Memory Management Overhead | CPU cost of ksmd, kswapd*, and kcompactd* |
Source Of Truth
The authoritative implementation for the observer lives under
cockpit/stakkr-observer/.
This page is the GitHub Pages navigation destination for it, not a second
implementation.