Add KubeAI monitoring support + vLLM dashboard #1054
Conversation
There was a problem hiding this comment.
Pull Request Overview
This pull request adds support for KubeAI monitoring and a vLLM dashboard for observability.
- Introduces a new YAML configuration file for Prometheus-based monitoring of vLLM metrics.
- Updates the README with instructions on enabling observability using the provided install script and Helm chart.
Reviewed Changes
Copilot reviewed 3 out of 4 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
| kubeai/metric-values.yaml | New configuration values enabling vLLM PodMonitor in Prometheus. |
| kubeai/README.md | Added Observability section with instructions for setting up monitoring and the vLLM dashboard. |
Files not reviewed (1)
- kubeai/install.sh: Language not supported
eero-t
force-pushed
the
kubeai-metrics
branch
2 times, most recently
from
cc4f8b8 to
7023f9e
Compare
Signed-off-by: Eero Tamminen <[email protected]>
In case somebody wants to run Helm directly instead of using install.sh. Signed-off-by: Eero Tamminen <[email protected]>
Signed-off-by: Eero Tamminen <[email protected]>
There was a problem hiding this comment.
Pull Request Overview
Adds Prometheus-based monitoring support for KubeAI’s vLLM engine and provides instructions for deploying a Grafana dashboard.
- Introduces
vLLMPodMonitorin Helm values for scraping vLLM metrics - Extends README with observability setup and dashboard installation steps
Reviewed Changes
Copilot reviewed 3 out of 4 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
| kubeai/metric-values.yaml | Enable Prometheus PodMonitor resource (vLLMPodMonitor.enabled) |
| kubeai/README.md | Add “Observability” section with script usage and dashboard setup |
Files not reviewed (1)
- kubeai/install.sh: Language not supported
Comments suppressed due to low confidence (1)
kubeai/README.md:170
- The new observability feature (Prometheus monitoring and vLLM dashboard) lacks automated tests. Consider adding unit or integration tests to validate the installation script and dashboard deployment.
# Observability
|
|
||
| metrics="" | ||
| for arg in "$@"; do | ||
| if [ -f "$arg" ]; then |
There was a problem hiding this comment.
nit: the arg parsing looks somewhat shaky'n'shady but I guess that's ok for this kind of hack/helper script
There was a problem hiding this comment.
I would think it very unlikely that user's Prometheus release is named exactly the same as some file in the kubeai/ directory...
PS. I'm wondering about the benefit of the script, when more things are needed to configure. I think it would be clearer if user would just invoke Helm directly (with command copy-pasted from README), in this case with additional -f monitoring.yaml argument.
3 participants
Description
Add KubeAI monitoring support + vLLM dashboard.
Monitoring can be added either by using the helper script, or by calling Helm directly with the new metrics.
Issues
n/a.Type of change
Dependencies
n/a.Tests
Manually tested.