2024-02-24 11:28:55 +00:00
|
|
|
# Server tests
|
|
|
|
|
|
2024-03-02 21:00:14 +00:00
|
|
|
Python based server tests scenario using [BDD](https://en.wikipedia.org/wiki/Behavior-driven_development)
|
|
|
|
|
and [behave](https://behave.readthedocs.io/en/latest/):
|
|
|
|
|
|
|
|
|
|
* [issues.feature](./features/issues.feature) Pending issues scenario
|
|
|
|
|
* [parallel.feature](./features/parallel.feature) Scenario involving multi slots and concurrent requests
|
|
|
|
|
* [security.feature](./features/security.feature) Security, CORS and API Key
|
|
|
|
|
* [server.feature](./features/server.feature) Server base scenario: completion, embedding, tokenization, etc...
|
2024-02-24 11:28:55 +00:00
|
|
|
|
|
|
|
|
Tests target GitHub workflows job runners with 4 vCPU.
|
|
|
|
|
|
2024-03-02 21:00:14 +00:00
|
|
|
Requests are
|
|
|
|
|
using [aiohttp](https://docs.aiohttp.org/en/stable/client_reference.html), [asyncio](https://docs.python.org/fr/3/library/asyncio.html)
|
|
|
|
|
based http client.
|
2024-02-24 11:28:55 +00:00
|
|
|
|
2024-03-02 21:00:14 +00:00
|
|
|
Note: If the host architecture inference speed is faster than GitHub runners one, parallel scenario may randomly fail.
|
|
|
|
|
To mitigate it, you can increase values in `n_predict`, `kv_size`.
|
2024-02-24 11:28:55 +00:00
|
|
|
|
|
|
|
|
### Install dependencies
|
2024-03-02 21:00:14 +00:00
|
|
|
|
2024-02-24 11:28:55 +00:00
|
|
|
`pip install -r requirements.txt`
|
|
|
|
|
|
|
|
|
|
### Run tests
|
2024-03-02 21:00:14 +00:00
|
|
|
|
2024-02-24 11:28:55 +00:00
|
|
|
1. Build the server
|
2024-03-02 21:00:14 +00:00
|
|
|
|
2024-02-24 11:28:55 +00:00
|
|
|
```shell
|
|
|
|
|
cd ../../..
|
2024-06-12 23:41:52 +00:00
|
|
|
cmake -B build -DLLAMA_CURL=ON
|
|
|
|
|
cmake --build build --target llama-server
|
2024-02-24 11:28:55 +00:00
|
|
|
```
|
2024-03-02 21:00:14 +00:00
|
|
|
|
|
|
|
|
2. Start the test: `./tests.sh`
|
2024-02-24 11:28:55 +00:00
|
|
|
|
|
|
|
|
It's possible to override some scenario steps values with environment variables:
|
2024-03-02 21:00:14 +00:00
|
|
|
|
|
|
|
|
| variable | description |
|
|
|
|
|
|--------------------------|------------------------------------------------------------------------------------------------|
|
|
|
|
|
| `PORT` | `context.server_port` to set the listening port of the server during scenario, default: `8080` |
|
2024-06-12 23:41:52 +00:00
|
|
|
| `LLAMA_SERVER_BIN_PATH` | to change the server binary path, default: `../../../build/bin/llama-server` |
|
2024-03-02 21:00:14 +00:00
|
|
|
| `DEBUG` | "ON" to enable steps and server verbose mode `--verbose` |
|
|
|
|
|
| `N_GPU_LAYERS` | number of model layers to offload to VRAM `-ngl --n-gpu-layers` |
|
2024-02-24 11:28:55 +00:00
|
|
|
|
|
|
|
|
### Run @bug, @wip or @wrong_usage annotated scenario
|
|
|
|
|
|
|
|
|
|
Feature or Scenario must be annotated with `@llama.cpp` to be included in the default scope.
|
2024-03-02 21:00:14 +00:00
|
|
|
|
2024-02-24 11:28:55 +00:00
|
|
|
- `@bug` annotation aims to link a scenario with a GitHub issue.
|
|
|
|
|
- `@wrong_usage` are meant to show user issue that are actually an expected behavior
|
|
|
|
|
- `@wip` to focus on a scenario working in progress
|
2024-03-02 21:00:14 +00:00
|
|
|
- `@slow` heavy test, disabled by default
|
2024-02-24 11:28:55 +00:00
|
|
|
|
|
|
|
|
To run a scenario annotated with `@bug`, start:
|
2024-03-02 21:00:14 +00:00
|
|
|
|
|
|
|
|
```shell
|
2024-03-17 18:12:37 +00:00
|
|
|
DEBUG=ON ./tests.sh --no-skipped --tags bug --stop
|
2024-03-02 21:00:14 +00:00
|
|
|
```
|
2024-02-24 11:28:55 +00:00
|
|
|
|
|
|
|
|
After changing logic in `steps.py`, ensure that `@bug` and `@wrong_usage` scenario are updated.
|
2024-03-02 21:00:14 +00:00
|
|
|
|
|
|
|
|
```shell
|
|
|
|
|
./tests.sh --no-skipped --tags bug,wrong_usage || echo "should failed but compile"
|
|
|
|
|
```
|
