Add AlmaLinux support

[eabdullin1]

Hello!

These changes add support for AlmaLinux.

We have successfully tested it on AlmaLinux versions 8 and 9, as well as CentOS 8 Stream and CentOS 9 Stream, and can confirm that everything is working as expected.

Going forward, our plan is to continue to contribute to this project. Specifically, we aim to add more tests to ensure robustness and reliability.

Thank you for your continued support.

[andrewlukoshko]

@carlosrodfern @AdamSaleh Could you please check this PR?

[bstinsonmhk]

Overall looks OK to me. I think in general, if we want t_functional to be useful in other places we may want to start factoring as much OS specific things into the initialization steps.

The Stream team may need to take a closer look to be sure this works properly in the CentOS Stream test harnesses.

[bstinsonmhk]

Can we use $os_name here instead of hard-coding the paths?

[eabdullin1]

Ok, I will try to use this way

[bstinsonmhk]

Can we factor these into their own vars in 0_lib?

[eabdullin1]

Do we really should put it to 0_lib?
I used it only for reusable vars, but this signing_key var is used only in this file

[bstinsonmhk]

Same thing here, we might want to factor the OS-related vars into their own groups in 0_lib

[eabdullin1]

Same as in previous

[AdamSaleh]

@eabdullin1 so, I have tested this on x86 vm on against stable composes of CentOS Stream 8 and 9, it looks it doesn't interfere with any of our tests, so if you'd do those few refactors Brian suggested, we should have no problems merging it.

I think, w.r.t. those vars that are specific to single files (key, singing_key, e.t.c), sometimes you can make test more generic, i.e. instead of

 lines_to_check="ALMALINUX_MANTISBT_PROJECT=\"AlmaLinux-$centos_ver\" ALMALINUX_MANTISBT_PROJECT_VERSION=\"$centos_ver.$minor_ver\""

you could use something like

lines_to_check="${os_name^^}_MANTISBT_PROJECT=\"$os_name-$centos_ver\" ${os_name^^}_MANTISBT_PROJECT_VERSION=\"$centos_ver.$minor_ver\"

For things that are harder to do like this, we would still prefer to have them in one place, so that we can see where we diverge and the rest of the tests are not populated by vendor-specific special cases.

We don't mind it is only used once. Still will prefer to see all of these somewhere top-level in the 0_lib/functions.sh file.

After that is done I would run it in more of our testing environments and after that works, I think we are good to go.

[eabdullin1]

Ok, thank you for the reply, will fix it

[carlosrodfern]

@AdamSaleh ,
The only test failing I saw was the firefox one. I opened an issue to address it in firefox itself: https://issues.redhat.com/browse/RHEL-2409. I provided MRs as well with a fix that uses the HOME_URL from /etc/os-release, so if it is accepted, that test will need to be updated. It could also remove the need of setting the website variable if Alma startup page is the same as the HOME_URL one.

[eabdullin1]

@AdamSaleh @carlosrodfern Guys, can you please take a look?

[carlosrodfern]

@eabdullin1 ,
A suggestion, how about removing the "is_almalinux" all together and instead use vars and flags set in a case/esac statement? I think this MR is very close to achieve that.

For example, in the functions.sh file, grab the OS from the ID variable in /etc/os-release, and then use it in a case to set variables and flags, for instance, to skip the z tests.

os_id=$(source /etc/os-release; echo $ID)

case $os_id in
	almalinux)
                # almalinux variables here
		should_skip_z_tests=1
		;;
	centos)
		# variables here
		;;
	*)
		# default variables here
		;;
esac

Then,

  if [ -z "$should_skip_z_tests" ]; then
    t_Process <(/usr/bin/find ./tests/z_*/ -type f|sort -t'/' )
  fi

... or something along those lines.

[eabdullin1]

@eabdullin1 , A suggestion, how about removing the "is_almalinux" all together and instead use vars and flags set in a case/esac statement? I think this MR is very close to achieve that.

For example, in the functions.sh file, grab the OS from the ID variable in /etc/os-release, and then use it in a case to set variables and flags, for instance, to skip the z tests.

os_id=$(source /etc/os-release; echo $ID)

case $os_id in
	almalinux)
                # almalinux variables here
		should_skip_z_tests=1
		;;
	centos)
		# variables here
		;;
	*)
		# default variables here
		;;
esac

Then,

  if [ -z "$should_skip_z_tests" ]; then
    t_Process <(/usr/bin/find ./tests/z_*/ -type f|sort -t'/' )
  fi

... or something along those lines.

@carlosrodfern
Using this way is not a problem, but I don't know what should be used as default.
Probably it will be better to use centos as default variables.

os_id=$(source /etc/os-release; echo $ID)

vendor="centos"
os_name="CentOS"
grub_sb_token='CentOS Secure Boot Signing 202'
kernel_sb_token="CentOS Secure Boot Signing 201"
key_template="CentOS \(Linux \)\?%s signing key"
firefox_start_page="www.centos.org"

case $os_id in
    almalinux)
        # AlmaLinux variables
        vendor="almalinux"
        os_name="AlmaLinux"
        grub_sb_token='AlmaLinux OS Foundation'
        kernel_sb_token=$grub_sb_token
        key_template="AlmaLinux %s signing key"
        firefox_start_page="www.almalinux.org"
        ;;
    *)
        # Keep default(CentOS) variables
        ;;
esac

[carlosrodfern]

@eabdullin1 ,
Yeah, that's fine. You could also have an entry for centos in the case, and if centos nor almalinux is found, simply exit 1 with "unsupported os" error. However, I defer to @AdamSaleh on that one since he is a collaborator.

More than the case/esac, the most important point is the removal of distro specific variable naming outside of functions.sh, to make the test suite as agnostic of EL distro as possible. I think it would be cleaner to remove the is_almalinux var and use instead flags that enable and disable things, e.g., should_skip_z_tests, in this specific case (which is the only one left):

  # For now we skipping these tests on AlmaLinux
  if [[ $is_almalinux == "no" ]]; then
    t_Process <(/usr/bin/find ./tests/z_*/ -type f|sort -t'/' )
  fi

[eabdullin1]

@carlosrodfern Sorry for the huge delay, I updated functions.sh, please review it)

[carlosrodfern]

@eabdullin1,
I just want to let you know that this conversation is also happening in parallel: https://discussion.fedoraproject.org/t/contributing-compose-tests-to-centos-stream

It will increase the ability of AlmaLinux to reuse tests and contribute if AlmaLinux migrates to tmt as well.