Skip to content

[Bugfix] Fix incorrect kv cache metrics in grafana.json #27133

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Oct 23, 2025
+6 −6
Merged

[Bugfix] Fix incorrect kv cache metrics in grafana.json #27133

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
f3c5ca8
fix: kv cache panel in grafana.json
Oct 17, 2025
8906e86
replace deprecated metrics under pserses
Oct 20, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion examples/online_serving/dashboards/perses/performance_statistics.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -530,7 +530,7 @@ spec:
name: accelerators-thanos-querier-datasource
# Multiply by 100 so we can read it as a percentage without setting a unit (avoids CUE unit conflicts)
query: >
100 * avg(vllm:gpu_cache_usage_perc)
100 * avg(vllm:kv_cache_usage_perc)

"18":
kind: Panel
Expand Down
8 changes: 4 additions & 4 deletions examples/online_serving/dashboards/perses/query_statistics.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -98,7 +98,7 @@ spec:
kind: PrometheusTimeSeriesQuery
spec:
datasource: { kind: PrometheusDatasource, name: accelerators-thanos-querier-datasource }
query: avg(vllm:gpu_cache_usage_perc{namespace="$NS",service="$SVC"}) or vector(0)
query: avg(vllm:kv_cache_usage_perc{namespace="$NS",service="$SVC"}) or vector(0)
minStep: "15s"

core_running_ts:
Expand Down Expand Up @@ -168,7 +168,7 @@ spec:
spec:
datasource: { kind: PrometheusDatasource, name: accelerators-thanos-querier-datasource }
# multiply by 100 to present percentage; omit format.unit to avoid schema conflicts
query: (avg(vllm:gpu_cache_usage_perc{namespace="$NS",service="$SVC"}) * 100) or vector(0)
query: (avg(vllm:kv_cache_usage_perc{namespace="$NS",service="$SVC"}) * 100) or vector(0)
minStep: "15s"

core_kv_usage_pct_ts:
Expand All @@ -187,7 +187,7 @@ spec:
kind: PrometheusTimeSeriesQuery
spec:
datasource: { kind: PrometheusDatasource, name: accelerators-thanos-querier-datasource }
query: (avg by (service) (vllm:gpu_cache_usage_perc{namespace="$NS",service="$SVC"}) * 100) or vector(0)
query: (avg by (service) (vllm:kv_cache_usage_perc{namespace="$NS",service="$SVC"}) * 100) or vector(0)
minStep: "15s"

# --- Per-Pod breakdowns (works on Simulator & Real) ---
Expand Down Expand Up @@ -246,7 +246,7 @@ spec:
spec:
datasource: { kind: PrometheusDatasource, name: accelerators-thanos-querier-datasource }
# if your exporter labels kv metric with pod (the sim does), this works; otherwise it will just return empty
query: (avg by (pod) (vllm:gpu_cache_usage_perc{namespace="$NS",service="$SVC"}) * 100) or vector(0)
query: (avg by (pod) (vllm:kv_cache_usage_perc{namespace="$NS",service="$SVC"}) * 100) or vector(0)
minStep: "15s"

# --- Real vLLM only (zeros on simulator) ---
Expand Down
2 changes: 1 addition & 1 deletion examples/online_serving/prometheus_grafana/grafana.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -852,7 +852,7 @@
"uid": "${DS_PROMETHEUS}"
},
"editorMode": "code",
"expr": "vllm:gpu_cache_usage_perc{model_name=\"$model_name\"}",
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

critical

The metric name vllm:gpu_cache_usage_perc is outdated and should be updated to vllm:kv_cache_usage_perc to align with the latest metrics. This discrepancy could lead to incorrect monitoring and alerting.

It's critical to ensure that monitoring dashboards use the correct metric names to provide accurate insights into system performance and resource utilization.

          "expr": "vllm:kv_cache_usage_perc{model_name=\"$model_name\"}"

"expr": "vllm:kv_cache_usage_perc{model_name=\"$model_name\"}",
"instant": false,
"legendFormat": "GPU Cache Usage",
"range": true,
Expand Down