MDBF-800 Extend package minor upgrade tests to ensure transition

[elenst]

… from old bb

Parts 1 and 2: Switching minor upgrade into "all" mode and extending it with additional checks (ldd, requirements, capabilities)

[elenst]

I'm not sure what I'm supposed to do about this "pre-commit" check. We do not need sudo to affect redirect, and replacing it with tee will only flood the output unnecessarily. For the remaining "not quoted" message, quoting it the way they suggest will make rpm command fail everywhere when the option is not usable and thus the variable remains empty.

[RazvanLiviuVarzaru]

I'm not sure what I'm supposed to do about this "pre-commit" check. We do not need sudo to affect redirect, and replacing it with tee will only flood the output unnecessarily. For the remaining "not quoted" message, quoting it the way they suggest will make rpm command fail everywhere when the option is not usable and thus the variable remains empty.

Hi, @elenst !

Thank you for your contribution! I'll try to review as much as I can on Monday, as I'll be on vacation for the rest of the week.

Just a couple of quick comments to help you resolve the pre-commit issues:

If you're confident that the inputs do not contain spaces, line feeds, glob characters, and similar elements, you can safely ignore SC2086 by adding the following comment above the affected line of code:

# shellcheck disable=SC2086

For redirection, the current user’s shell will be used to write the output to a file. If there are no permission issues at this point, you can safely ignore the warning by adding:

# shellcheck disable=SC2024

Maybe @fauust can have o a quick look on the scripts too.

[elenst]

you can safely ignore the warning

Thanks, done (kind of, at least I've made it pass).

[fauust]

Thanks @elenst for implementing those extra missing checks, it was planned but I did not found time to implement and was not sure how to proceed (MDBF-401).

Overall this looks good but it's not working ATM, see my comments. BTW if you want to test those scripts manually, you can call the script on a clean VM with something like:
./rpm-upgrade.sh https://buildbot.mariadb.org/#/builders/608/builds/1759 (only jq sudo wget are required and a sudo NOPASSWD user).

[fauust]

- bb_log_error "ERROR: found changes in server capabilities"
+ bb_log_err "found changes in server capabilities"

[fauust]

sed 's/bb_log_error/bb_log_err/g'

[elenst]

fixed.

if you want to test those scripts manually, you can call the script on a clean VM

That's a great capability, helps a lot, thanks.
Checked:

• minor and major upgrade on alma8-amd64 10.5 (minor upgrade fails, but on the "right" reasons, due to differences in dependencies and requirements)
• minor and major upgrade on ubuntu2004-amd64 10.5 (it passes as far as scripts are concerned, but I noticed that the check for mysql_upgrade is fragile, on my very slow VM the test failed several times as it just didn't wait long enough. We'll need to see if it happens in the buildbot, but since it's an old check, it is out of the scope of this PR)
• minor upgrade on fedora39-amd64 10.6 (failed due to a requirement difference)

[fauust]

Good! Regarding https://github.com/koalaman/shellcheck/wiki/SC2024, agree, it's a bit strange and we might consider ignore it everywhere if that helps, shellchek is usually very useful but sometimes also wrong :)

minor and major upgrade on ubuntu2004-amd64 10.5 (it passes as far as scripts are concerned, but I noticed that the check for mysql_upgrade is fragile, on my very slow VM the test failed several times as it just didn't wait long enough. We'll need to see if it happens in the buildbot, but since it's an old check, it is out of the scope of this PR)

Let's see how it goes after the merge, I have triggered a 10.5 and 10.11 rebuild on DEV. When @RazvanLiviuVarzaru is back from vacation (next week), we need to decide how we can make these triggers available to everyone on DEV (ATM, triggered by any push on my fork).