... | ... | @@ -41,6 +41,32 @@ Those configurations are run for each merge request events and on push to the (u |
|
|
* Image: `medialibrary-win64`
|
|
|
* All unit & functional tests are run
|
|
|
|
|
|
## Automated requests benchmarking
|
|
|
|
|
|
For merge requests tagged with ~"model-change" 2 additional jobs will be run:
|
|
|
|
|
|
### gen-benchmark-db
|
|
|
|
|
|
This job will generate a new benchmark database, to be used in the `run-bench` job.
|
|
|
|
|
|
It will start by generating 100 artists with 10 albums of 10 tracks each. Those samples will be discovered to generate a database that will be queried by the benchmarks.
|
|
|
|
|
|
This job will generate an artifact named `bench.<MILESTONE>.db` where `<MILESTONE>` is the milestone targeted by the merge request.
|
|
|
|
|
|
If the job is run manually, the targetted branch will be used in place of `<MILESTONE>`.
|
|
|
|
|
|
The resulting artifact will also be stored in https://artifacts.videolan.org/medialibrary/gen-bench-db/
|
|
|
|
|
|
### run-bench
|
|
|
|
|
|
This job will run the [requests benchmark](https://code.videolan.org/videolan/medialibrary/-/tree/master/benchmark/requests). Each specific benchmark will be run 10 times.
|
|
|
|
|
|
The result is then compared to the previous run for this database model (if any).
|
|
|
|
|
|
An external job will be added with the results. If the performance delta is larger than 5% it will be interpreted as a meaningful performance increase/decrease.
|
|
|
|
|
|
This is currently not super helpful as the results will depend on the CI machine load.
|
|
|
|
|
|
## Scheduled jobs
|
|
|
|
|
|
### Nightly jobs
|
... | ... | |