[WIP] feat: use runners to startup the services

[jvillafanez]

Description

Replace oklog library with customized runners. This is intended to simplify and homogenize command's behavior. Some of the os.Exit calls will also be removed.

Note: It's expected that all the services will follow the same approach. The only notable exception (for now) is the antivirus service, which can't be stopped. This will need further analysis.

Related Issue

• Fixes <issue_link>

Motivation and Context

Runner's behaviour is expected to be the same across all the commands. The code is also simplified because the behavior is set within the runner package, so the commands just need to provide the parameters.

How Has This Been Tested?

• test environment:
• test case 1:
• test case 2:
• ...

Screenshots (if appropriate):

Types of changes

• [ ] Bug fix (non-breaking change which fixes an issue)
• [ ] New feature (non-breaking change which adds functionality)
• [ ] Breaking change (fix or feature that would cause existing functionality to change)
• [ ] Technical debt
• [ ] Tests only (no source changes)

Checklist:

• [ ] Code changes
• [ ] Unit tests added
• [ ] Acceptance tests added
• [ ] Documentation ticket raised:

[update-docs[bot]]

Thanks for opening this pull request! The maintainers of this repository would appreciate it if you would create a changelog item based on your changes.

[jvillafanez]

We'll need at least an additional factory method for the reva runtime services. This will affect the following services:

• users
• storage-*
• sharing
• ocm
• groups
• gateway
• frontend
• auth-*
• app-registry
• app-provider

This will need some investigation about how to properly stop the runner for those services.

In addition, there are other services that might need special handling:

• postprocessing: we could create an ad-hoc runner for this. Solution shouldn't differ greatly of what we have at the moment, maybe just adding another way out in case the stopper is called. (https://github.com/owncloud/ocis/issues/9096)
• policies
• ocdav
• notifications
• nats
• idm
• clientlog
• audit
• antivirus: investigation is needed because it seems we can't stop the service. We might need changes in the code

[jvillafanez]

It seems I'm stuck with the reva-related services. I think we need heavy changes in the reva runtime in order to keep going.

The reva runtime has multiple problems:

• No clear Stop method that could be use to stop the service. The method seems to be missing, so it doesn't seem possible to request the service to stop.
• Handling OS signals by itself. As a service, reva shouldn't care about the OS signal. It should provide the means to manage the service and let the app (oCIS in this case) to control the signals the way it wants in order to provide the oCIS' desired behavior. For example, oCIS could intercept a "quit" signal in order to provide a controlled and predictable shutdown process. This could involve, for example, saving important data in the storage, or letting upload and download requests to finish (which involves saving data in the storage). However, currently, reva will shutdown anyway, and could break ongoing uploads or downloads.
• There are multiple os.Exit calls which will cause the whole app to close. A graceful shutdown is impossible.

As a possible alternative, we could try to spawn a new process for each service. However, the solution also have some problems.

• We'd need the location of the reva executable (or figure out where it is) in order to start a new process.
• There are quite a bunch of reva services that would need to be launched. This means, a lot of processes.
• We'd need to manage several processes. This might become problematic, and we'll likely need to implement additional safety measures in case something goes wrong (a process could not be stopped and relaunching it after restarting oCIS could cause problems)

It seems a better idea to wait for changes in reva so it's easier to handle its services.

[sonarqubecloud[bot]]

Quality Gate passed

Issues
14 New issues
0 Accepted issues

Measures
0 Security Hotspots
2.2% Coverage on New Code
37.9% Duplication on New Code

See analysis details on SonarCloud