follow-up #10 (which was a git hickup)
## Behaviour
Log purging is attempted in an asynchronous process before a new pipeline of a specific repo is started.
It does so by
1. Getting all existing pipelines
2. Filtering by `keepMin` and `keepDuration` settings
3. Calling `LogDelete` for all remaining pipelines
Deleting only logs instead of the full pipeline (which `crow-cli pipeline purge` does) is preferred to keep historic pipeline information. Storing this in the DB is just a single line and doesn't contain much content (in contrast to logs).
## Defaults
- No minimum count is kept (`CROW_DEFAULT_LOGS_PIPELINES_KEEP_MIN`)
- All pipelines of the last 90 days (per repo) are kept (`CROW_DEFAULT_LOGS_KEEP_DURATION`)
## Todo
- [x] implement purge call during pipeline start
- [x] add settings to DB column and repo settings
- [x] tests
- [x] think about defaults
- [x] Currently the purge happens on all pipelines in scope, including ones which have already been cleared. To avoid these unnecessary calls, which also will add up for repos with many pipelines, an indicator is needed which allows filtering these pipelines out.
fix#9
Co-authored-by: crowci-bot <admin@crowci.dev>
Reviewed-on: https://codeberg.org/crowci/crow/pulls/34
Co-authored-by: pat-s <patrick.schratz@gmail.com>
Co-committed-by: pat-s <patrick.schratz@gmail.com>
Try to fix#1495
It's very hard to reproduce it and only way to fix when it gets in this
state is woodpecker agent restart.
This anyway fixes problem if step mounts and
`WOODPECKER_BACKEND_DOCKER_VOLUMES` conflict
