Fix: add init log option for velero controller-runtime manager.

[blackpiglet]

Signed-off-by: Xun Jiang jxun@vmware.com

Thank you for contributing to Velero!

Please add a summary of your change

Does your change fix a particular issue?

Fixes #3737

Please indicate you've done the following:

• Accepted the DCO. Commits without the DCO will delay acceptance.
• Created a changelog file or added /kind changelog-not-required.
• Updated the corresponding documentation in site/content/docs/main.

[jenting]

In general, LGTM
But it's better to have a changelog if it's fixing an issue.

[blackpiglet]

@jenting
Thanks. I will update the changelog.

[ywk253100]

@blackpiglet Could we try to address the issues mentioned here? If we simply set a zap logger for the controller, there will be two kinds of log formats. And we also need to support setting the log level for the new logger. Implement the logger interface with logrus is one solution

[blackpiglet]

@ywk253100

I checked whether it's possible to use logrus logger, but it seems the manager only takes logger implements the github.com/go-logr/logr interface. logrus and logr's interface are not compatible.

Logger logr.Logger

[sseago]

@blackpiglet check out "github.com/bombsimon/logrusr" -- it allows you to implement the go-logr logr.Logger interface with a logrus logger.

[sseago]

@blackpiglet For the openshift velero plugin used with OADP, we needed to call a shared func that was used both in the plugin (which uses velero's logrus implementation) and in the crane migration tool (which uses go-logr Loggers).

This line calls logrusr.NewLogger(p.Log), passing in a logrus.FieldLogger, and getting a logr.Logger
https://github.com/openshift/openshift-velero-plugin/blob/master/velero-plugins/imagestream/backup.go#L88

The func we're calling there takes a logr.Logger param rather than a logrus.FieldLogger

[blackpiglet]

@sseago
Thanks a lot. Your suggestion really saves a lot of my work!
As a matter of fact, I tried to write a log adaptor by myself yesterday. Although it worked, there are still a lot of things I'm clear about in the code.

[ywk253100]

@blackpiglet And one thing we need to do is to remove the similar code like this which will output duplicated logs after we set logger for the controller.

We can do this during the controller refactoring from kubebuilder v2 to v3, but please open a new issue for this if you don't include the change in this PR

[blackpiglet]

@ywk253100
Great catch. After adding the new logger, it may introduce duplicated log entries.
I opened a new issue to trace the potential duplicated logs. #4368