Add parameter to skip ignoring files

[apilaskowski]

In our use-case we have a race condition.

We are creating files, then we are setting up privileges. In function 'extend_paths' we can see that if file is not readable once it is ignored, so it happened to us, that some logs were ignored forever.

I think that introducing such parameter should be the easiest way to avoid similar problems in the future.

Which issue(s) this PR fixes:
Fixes #
Rare race condition with updating privileges to files after executing 'in_tail.rb'

What this PR does / why we need it:

Docs Changes:
New parameter in 'in_tail'

Release Note:

[apilaskowski]

Why is this DCO failing? I added message required by it at the end of the commit. Can someone help?
Also, should I prepare documentation changes in this PR also, or should it be done separately?

[ashie]

Thanks for your contribution!

Why is this DCO failing?

The answer is described in the link:

Expected "Artur Piotr Izaak Laskowski apilaskowski@google.com", but got "Artur Laskowski apilaskowski@google.com".

[apilaskowski]

DCO is now passing

[apilaskowski]

I will try to fix it later today, I assume this is some problem with new parameter not being properly created. I will take a look

[apilaskowski]

@ashie PTAL, i made a misspelling before

[daipom]

Thanks for requesting the new feature!

that some logs were ignored forever.

Does this mean that the following warning log will be no longer shown?

fluentd/lib/fluent/plugin/in_tail.rb

Lines 306 to 312 in 95438b2

	if is_file
	unless @ignore_list.include?(p)
	log.warn "#{p} unreadable. It is excluded and would be examined next time."
	@ignore_list << p if @ignore_repeated_permission_error
	end
	end
	false

Or mean that the collection of the file will stop forever?

[apilaskowski]

What I want to achieve is not ignoring any files, it happens that they are unreadable at the beginning, but then gets readable after some time.

I am wondering and probably missed that at the beginning, is ignore_repeated_permission_error doing just that?
I will investigate and get back to you.

[apilaskowski]

I think that ignore_repeated_permission_error is working for me, thank you for your cooperation.
If I will need more I will return to this, however, for now it is working for us.

[daipom]

@apilaskowski
Thanks for your feedback!

What I want to achieve is not ignoring any files, it happens that they are unreadable at the beginning, but then gets readable after some time.

@ignore_list is just a list to control the warning message.
The list has nothing to do with whether the file is collected or not.
If a file is added to the list, the file should be collected after it gets readable.

If the file is not collected after it gets readable, there may be some other issues.
If you have any problems collecting files, please tell us again!