[READ-ONLY] Mirror of https://github.com/FoxxMD/crowdsecBench. Benchmark Crowdsec CPU usage with concurrent requests
benchmark
cpu-usage
crowdsec
docker
k6
monitoring
visualization
refactor: Remove env requirement
* default url in script so not env is necessary
* make env_file optional in compose
* Use defaults in script, or load env, for logging and generating goeffel graph subtitle
+30
-20
+13
-14
README.md
+13
-14
README.md
···
16
16
17
17
Additionally, the bash script [`bench.sh`](/bench.sh) can be used to orchestrate the load call, monitor the crowdsec process, and create a .png plot of CPU using [`goeffel`](https://github.com/jgehrcke/goeffel).
18
18
19
-
# Usage
20
-
21
-
## Create .env
19
+
## Usage
22
20
23
-
Use the sample [`.env.example`](/.env.example) to create an `.env` for your scenario.
24
-
25
-
* `TARGET` = number of concurrent requests to make per second
26
-
* `DURATION` = number of seconds to run load
27
-
* `RANDOMIZE` = randomize url path
28
-
* `URL` = leave as default unless modifying this project for your own use
21
+
### Run Load
29
22
30
-
## Run Load
31
-
32
-
### Using `bench.sh`
23
+
#### Using `bench.sh`
33
24
34
25
```shell
35
26
./bench.sh
···
39
30
40
31
Additional graphs can be plotted from the generated `hdf5` file using [`goeffel-analysis`](https://github.com/jgehrcke/goeffel?tab=readme-ov-file#goeffel-analysis-data-inspection-and-visualization).
41
32
42
-
### Manually
33
+
#### Manually
43
34
44
35
1. Start docker services required to run the load
45
36
···
61
52
docker compose up k6
62
53
```
63
54
64
-
# Benchmark Results
55
+
### (Optional) Customize Load with .env
56
+
57
+
Use the sample [`.env.example`](/.env.example) to create an `.env` for your scenario.
58
+
59
+
* `TARGET` = number of concurrent requests to make per second
60
+
* `DURATION` = number of seconds to run load (the `s` at the end is important)
61
+
* `RANDOMIZE` = randomize url path -- `true` or `false`
62
+
63
+
## Benchmark Results
65
64
66
65
See the [`bench-results`](/bench-results/) folder for some benchmarks run on my own hardware.
+12
-2
bench.sh
+12
-2
bench.sh
···
2
2
3
3
trap 'exit 0;' SIGHUP SIGINT
4
4
5
+
TARGET='50'
6
+
DURATION='30s'
7
+
8
+
if test -f ".env"; then
9
+
printf 'Found .env\n'
10
+
export $(cat .env | xargs)
11
+
fi
12
+
13
+
printf 'RUN => %s concurrent requests over %s\n' "$TARGET" "$DURATION"
14
+
5
15
printf 'Starting traefik and echo...\n'
6
16
docker compose up traefik echo -d
7
17
···
21
31
printf 'Starting goeffel capture of crowdsec PID %s \n' "$CS_PID"
22
32
docker compose run --rm --no-TTY goeffel goeffel --pid $CS_PID &
23
33
24
-
printf 'Starting k6\n'
34
+
printf 'Starting k6, will run for %s\n' "$DURATION"
25
35
docker compose up k6 -d
26
36
27
37
docker compose wait k6
···
44
54
--column proc_cpu_util_percent_total \
45
55
'CPU util (total) / %' \
46
56
"Crowdsec Web Traffic Load $timestamp" 5 \
47
-
--subtitle '50 concurrent requests, measured with Goeffel' \
57
+
--subtitle "$TARGET concurrent requests, measured with Goeffel" \
48
58
--legend-loc 'upper right' \
49
59
--mean-style solid \
50
60
--custom-y-limit 0 100
+3
-1
compose.yaml
+3
-1
compose.yaml
+2
-2
src/k6/script.ts
+2
-2
src/k6/script.ts
···
12
12
import {generateRandomUserAgent} from './utils/agent.ts';
13
13
14
14
const duration = __ENV.DURATION ?? '30s';
15
-
let target = __ENV.TARGET ?? 2;
15
+
let target = __ENV.TARGET ?? 50;
16
16
if(typeof target === 'string') {
17
17
target = parseInt(target);
18
18
}
···
32
32
}
33
33
};
34
34
35
-
const URL = __ENV.URL;
35
+
const URL = __ENV.URL ?? 'http://traefik';
36
36
37
37
const randomizeUrl = __ENV.RANDOMIZE === 'true';
38
38