Clear the data/log for jobs in Gitea Actions that are too old

[wolfogre]

Feature Description

Jobs that have been done for too long no longer need to retain data/log, so it is necessary to regularly clean them up.

• Log file: Delete Actions logs older than 1 year by default #31735
• Outputs.
• Artifacts: Artifacts retention and auto clean up #26131
• Old raws in job/run/task tables
• And more?

Details

#26275 (comment) : Some logic which could help:

• What you want delete is a run, so you need to delete a row of ActionRun.
• A run could include multiple jobs, so you need to delete multiple rows of ActionRunJob.
• When a job has been picked up by a runner to execute, there will be a task, however, a job could be rerun many times, so you need to delete multiple rows of ActionTask.
• Every task could include
  • multiple steps, so ActionTaskStep
  • multiple outputs, so ActionTaskOutput
• The log of the task is stored as a file (in DBFS or storage), so you need to detele it according to ActionTask.LogFilename or ActionTask.LogInStorage.
• In addition, there may be some artifacts uploaded to the run, so you need to delete multiple rows of ActionArtifact.
  • The files of artifacts are stored in storage, so you need to remove them.
• Finally, update NumActionRuns and NumClosedActionRuns of Repository.

[yp05327]

Maybe a related question?
To save the storage usages, is it worth compressing these data/log?

[ghnp5]

I was now looking for this 😊
Thanks!!

[lunny]

Maybe we can compress as zip, so that when display via HTTP, it can be unzip by web browser.

[ghnp5]

Maybe, but I think most important is to be able to purge too, with some Retention setting (e.g. older than 1 month, keep last 3, etc), so that the storage doesn't grow forever.

[nicorac]

+1 for purge ability; this is my use case:

I'm managing an internal Gitea server, and I'm going to make Actions available to my co-worker devs.

Since there's no other way to test push Actions than... pushing, I've played with them by "trial and error" 😉.
My tests caused the repo to actually contain 150+ completely useless job executions, so I'd like to purge them all before going to "production".

To make it worst, I suppose that once Actions will be made available to anyone on my server, my co-workers will also start testing them the same way... with the same result.

And finally, once in production, there's no need for us to keep any of repos Actions logs longer then 2wks.

[nodiscc]

Related #26219 (Delete workflow runs) (manually)

[diamante0018]

about the Artifacts would it be possible to add a little rubbish icon next to the file name so they can be deleted manually? (talking about artifacts created by actions and uploaded via actions/upload-artifact)

[bilogic]

For me, 7 days is more than enough

[f1mishutka]

For me, 7 days is more than enough

Would be much better to have this interval configurable per repository.

[bilogic]

After deleting the old action logs, the spinners don't stop and there are no messages to indicate the log no longer exists.