blob: 5416726a0e1d9c2f11ebce846224c3099ba6b797 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
|
# Dtail Metrics Example
This page walks through using Epimetheus watch mode with a CSV that could come from a tool like [Dtail](https://dtail.dev/) or any similar log/aggregation export.
## Scenario
You have a CSV file (e.g. `dtail.csv`) with columns that mix numeric stats and identifiers (host, service, etc.). You want to turn those into Prometheus metrics so you can graph them in Grafana.
## Steps
1. **Ensure the CSV has a header row**
First line = column names. Epimetheus will sanitize them for use as metric names and labels.
2. **Identify numeric vs string columns**
- Numeric columns (e.g. `count`, `avg_latency_ms`, `p99`) become metric values.
- String columns (e.g. `host`, `service`, `region`) become labels.
3. **Run watch mode** with a base metric name and your Prometheus (or Prometheus-compatible) write URL:
```bash
./epimetheus -mode=watch \
-file=dtail.csv \
-metric-name=dtail \
-prometheus=http://localhost:9090/api/v1/write
```
4. **Optional: resolve IPs to hostnames**
If one of your label columns contains IPs (e.g. `host` or `ip`), you can resolve them:
```bash
./epimetheus -mode=watch \
-file=dtail.csv \
-metric-name=dtail \
-prometheus=http://localhost:9090/api/v1/write \
-resolve-ip-labels=host
```
5. **Query in Prometheus / Grafana**
Metrics will appear as `dtail_<column_name>` with your string columns as labels, e.g.:
```promql
dtail_avg_latency_ms{service="api", region="eu"}
```
## References
- [CSV Format Flexibility](csv-format-flexibility.md) – how column types and names are interpreted.
- [DNS Resolution](dns-resolution.md) – IP-to-hostname resolution.
- [Operating Modes](modes.md) – all watch mode options.
|
