Extended test suit failure: `install_opam_bionic`, `documentation:linkcheck`, `install_opam_focal` (2022-06-17 -- ...)

The scheduled, extended test suite pipeline of 2022-06-17 failed for the jobs install_opam_bionic, documentation:linkcheck, install_opam_focal (2022-06-17 -- ...). See for more details: https://gitlab.com/tezos/tezos/-/pipelines/566131974/failures

Found 3 failures in pipeline #566131974 in project 3836952:

• job #2603259062 (install_opam_focal):

  To retry the full job, do: ./docs/scripts/test_doc_scripts.sh install-opam-focal

• job #2603259061 (install_opam_bionic):

  To retry the full job, do: ./docs/scripts/test_doc_scripts.sh install-opam-bionic

• job #2603259036 (documentation:linkcheck):

  To retry the full job, do: make -C docs linkcheck

What to do when a test in the extended test suite fails?

• The organizer of the test effort is responsible for looking at the result of scheduled pipelines (that @arvidnl = me)
  • In their absence, they'll find a replacement
• If a test fails, they'll perform a bisect, finding the merge on master for which the test first failed (the culprit commit).
  • Initially, this is manual for now but can be automated
• They create an issue (this one), tag it incident, assigning the author(s) of the culprit. I'll also ping the author of the test in question (if an obvious such author exists).
  • Being an incident, the culprit author should look at the issue ASAP and divert time to it. It should be decided whether:
    • if the fix can be merged the same day: then the test should be fixed, or
    • if the fix takes more time: the culprit should be reverted, unless
    • the culprit is an urgent important feature: then the test should be unplugged and a plan should be made for it's fix, typically in another part of the protocol cycle?

Discussion

This is a test run of the procedure above. Comment to describe your feelings about the whole affair.