Skip to content

chore(monitoring): shorten grafana query windows #81

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
GentleCold merged 1 commit into from
Jun 13, 2026
Merged

chore(monitoring): shorten grafana query windows #81

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
32 changes: 16 additions & 16 deletions deploy/monitoring/grafana/dashboards/daser-overview.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -252,7 +252,7 @@
},
"targets": [
{
"expr": "(sum(rate(daser_cache_lookup_total{job=\"daser\", result=\"hit\"}[5m])) or vector(0)) / clamp_min(sum(rate(daser_cache_lookup_total{job=\"daser\"}[5m])), 1)",
"expr": "(sum(rate(daser_cache_lookup_total{job=\"daser\", result=\"hit\"}[30s])) or vector(0)) / clamp_min(sum(rate(daser_cache_lookup_total{job=\"daser\"}[30s])), 1)",
"legendFormat": "request"
}
],
Expand Down Expand Up @@ -316,7 +316,7 @@
},
"targets": [
{
"expr": "rate(daser_cache_matched_tokens_total{job=\"daser\"}[5m]) / clamp_min(rate(daser_cache_requested_tokens_total{job=\"daser\"}[5m]), 1)",
"expr": "rate(daser_cache_matched_tokens_total{job=\"daser\"}[30s]) / clamp_min(rate(daser_cache_requested_tokens_total{job=\"daser\"}[30s]), 1)",
"legendFormat": "token"
}
],
Expand Down Expand Up @@ -470,7 +470,7 @@
},
"targets": [
{
"expr": "sum by (result) (rate(daser_cache_lookup_total{job=\"daser\"}[5m]))",
"expr": "sum by (result) (rate(daser_cache_lookup_total{job=\"daser\"}[30s]))",
"legendFormat": "{{result}}"
}
],
Expand Down Expand Up @@ -563,15 +563,15 @@
},
"targets": [
{
"expr": "histogram_quantile(0.5, rate(daser_cache_prefix_reuse_tokens_bucket{job=\"daser\"}[5m]))",
"expr": "histogram_quantile(0.5, rate(daser_cache_prefix_reuse_tokens_bucket{job=\"daser\"}[30s]))",
"legendFormat": "p50"
},
{
"expr": "histogram_quantile(0.95, rate(daser_cache_prefix_reuse_tokens_bucket{job=\"daser\"}[5m]))",
"expr": "histogram_quantile(0.95, rate(daser_cache_prefix_reuse_tokens_bucket{job=\"daser\"}[30s]))",
"legendFormat": "p95"
},
{
"expr": "histogram_quantile(0.99, rate(daser_cache_prefix_reuse_tokens_bucket{job=\"daser\"}[5m]))",
"expr": "histogram_quantile(0.99, rate(daser_cache_prefix_reuse_tokens_bucket{job=\"daser\"}[30s]))",
"legendFormat": "p99"
}
],
Expand Down Expand Up @@ -674,11 +674,11 @@
},
"targets": [
{
"expr": "increase(daser_cache_late_evicted_commits_total{job=\"daser\"}[5m])",
"expr": "increase(daser_cache_late_evicted_commits_total{job=\"daser\"}[30s])",
"legendFormat": "late commits"
},
{
"expr": "sum by (reason) (rate(daser_cache_evicted_chunks_total{job=\"daser\"}[5m]))",
"expr": "sum by (reason) (rate(daser_cache_evicted_chunks_total{job=\"daser\"}[30s]))",
"legendFormat": "evict {{reason}}"
}
],
Expand Down Expand Up @@ -766,11 +766,11 @@
},
"targets": [
{
"expr": "histogram_quantile(0.5, sum by (le, op) (rate(daser_transfer_duration_seconds_bucket{job=\"daser\"}[5m]))) * 1000",
"expr": "histogram_quantile(0.5, sum by (le, op) (rate(daser_transfer_duration_seconds_bucket{job=\"daser\"}[30s]))) * 1000",
"legendFormat": "p50 {{op}}"
},
{
"expr": "histogram_quantile(0.95, sum by (le, op) (rate(daser_transfer_duration_seconds_bucket{job=\"daser\"}[5m]))) * 1000",
"expr": "histogram_quantile(0.95, sum by (le, op) (rate(daser_transfer_duration_seconds_bucket{job=\"daser\"}[30s]))) * 1000",
"legendFormat": "p95 {{op}}"
}
],
Expand Down Expand Up @@ -851,7 +851,7 @@
},
"targets": [
{
"expr": "sum by (op) (rate(daser_transfer_bytes_total{job=\"daser\"}[5m])) / 1e9",
"expr": "sum by (op) (rate(daser_transfer_bytes_total{job=\"daser\"}[30s])) / 1e9",
"legendFormat": "{{op}}"
}
],
Expand Down Expand Up @@ -898,11 +898,11 @@
},
"targets": [
{
"expr": "histogram_quantile(0.5, sum by (le, op) (rate(daser_transfer_chunk_size_bytes_bucket{job=\"daser\"}[5m])))",
"expr": "histogram_quantile(0.5, sum by (le, op) (rate(daser_transfer_chunk_size_bytes_bucket{job=\"daser\"}[30s])))",
"legendFormat": "p50 {{op}}"
},
{
"expr": "histogram_quantile(0.95, sum by (le, op) (rate(daser_transfer_chunk_size_bytes_bucket{job=\"daser\"}[5m])))",
"expr": "histogram_quantile(0.95, sum by (le, op) (rate(daser_transfer_chunk_size_bytes_bucket{job=\"daser\"}[30s])))",
"legendFormat": "p95 {{op}}"
}
],
Expand Down Expand Up @@ -968,7 +968,7 @@
},
"targets": [
{
"expr": "max_over_time(((sum(increase(daser_l1_hits_total{job=\"daser\"}[1m])) or vector(0)) / clamp_min((sum(increase(daser_l1_hits_total{job=\"daser\"}[1m])) or vector(0)) + (sum(increase(daser_l1_misses_total{job=\"daser\"}[1m])) or vector(0)), 1))[2m:1s])",
"expr": "max_over_time(((sum(increase(daser_l1_hits_total{job=\"daser\"}[30s])) or vector(0)) / clamp_min((sum(increase(daser_l1_hits_total{job=\"daser\"}[30s])) or vector(0)) + (sum(increase(daser_l1_misses_total{job=\"daser\"}[30s])) or vector(0)), 1))[30s:1s])",
"legendFormat": "L1 hit rate"
}
],
Expand Down Expand Up @@ -1038,7 +1038,7 @@
},
"targets": [
{
"expr": "max_over_time(((sum(daser_l1_bytes_used{job=\"daser\"}) or vector(0)) / clamp_min((sum(daser_l1_bytes_capacity{job=\"daser\"}) or vector(0)), 1))[2m:1s])",
"expr": "max_over_time(((sum(daser_l1_bytes_used{job=\"daser\"}) or vector(0)) / clamp_min((sum(daser_l1_bytes_capacity{job=\"daser\"}) or vector(0)), 1))[30s:1s])",
"legendFormat": "L1 used"
}
],
Expand Down Expand Up @@ -1299,7 +1299,7 @@
},
"targets": [
{
"expr": "max_over_time(((sum(increase(vllm:request_success_total{job=\"vllm\"}[1m])) or vector(0)) / 60)[2m:1s])",
"expr": "max_over_time(((sum(increase(vllm:request_success_total{job=\"vllm\"}[30s])) or vector(0)) / 30)[30s:1s])",
"legendFormat": "requests/s"
}
],
Expand Down
6 changes: 3 additions & 3 deletions docs/monitoring/prometheus.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -143,9 +143,9 @@ job="vllm" — vLLM inference server
Use token hit ratio rather than only request hit ratio when judging benefit:

```promql
rate(daser_cache_matched_tokens_total[5m])
rate(daser_cache_matched_tokens_total[30s])
/
clamp_min(rate(daser_cache_requested_tokens_total[5m]), 1)
clamp_min(rate(daser_cache_requested_tokens_total[30s]), 1)
```

### Transfer & Storage
Expand All @@ -170,7 +170,7 @@ The IPC server also logs transfer summaries with decimal GB/s throughput:
Grafana should compute throughput from bytes:

```promql
rate(daser_transfer_bytes_total[5m]) / 1e9
rate(daser_transfer_bytes_total[30s]) / 1e9
```

### vLLM Inference (from vLLM scrape target)
Expand Down
Loading
Loading