[DPE-7584] Update single kernel library to fix tmpfs ownership and permission after reboot

[marceloneppel]

Issue

If temp storage is using tmpfs and the unit is rebooted, that storage is remounted with wrong permissions, leading to failures when a database user tries to create temporary objects in the database.

Solution

Update the single kernel library dependency to fix the issue.

Add an integration test for checking that the tmpfs storage works after a reboot.

Depends on canonical/postgresql-single-kernel-library#11.

Checklist

• I have added or updated any relevant documentation.
• I have cleaned any remaining cloud resources from my accounts.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 64.51%. Comparing base (798e91f) to head (fd7bd63).
⚠️ Report is 2 commits behind head on 16/edge.

❌ Your project status has failed because the head coverage (64.51%) is below the target coverage (70.00%). You can increase the head coverage or adjust the target coverage.

Additional details and impacted files

@@           Coverage Diff            @@
##           16/edge    #1156   +/-   ##
========================================
  Coverage    64.51%   64.51%           
========================================
  Files           17       17           
  Lines         4337     4337           
  Branches       671      671           
========================================
  Hits          2798     2798           
  Misses        1357     1357           
  Partials       182      182           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[marceloneppel]

Need to be changed to 16.0.1 after canonical/postgresql-single-kernel-library#11 is merged and published.

[taurus-forever]

Why tests are so unstable (again?). 16/edge suposed to be more stable. Nope?

[marceloneppel]

Why tests are so unstable (again?). 16/edge suposed to be more stable. Nope?

It's supposed. What's happening is the following: after a restore, the permissions of the storage directories are being set to a different value than the one that is set when the workload starts the database in a fresh cluster. I'm checking if there is a reason for that; otherwise, I'll fix those too.

[marceloneppel]

Why tests are so unstable (again?). 16/edge suposed to be more stable. Nope?

The issue is described at canonical/postgresql-single-kernel-library#17 description.

I fixed the issue in that PR (and imported the code of the commit on this PR just to ensure it really fixes the issue).

I also fixed the directory permissions used in the configure_patroni_on_unit method to match what we expect and avoid more issues when restarting a unit. The remaining places where some permissions diverge need to be checked and fixed in a follow-up PR.

[taurus-forever]

LGTM, should you use proper lib version?

[taurus-forever]

16.0.2?

[marceloneppel]

Exactly. Updated on fd7bd63.