ceph 18.2.2-3

[bazaah]

• Backport ceph/ceph@9ee47b0 (fix for ceph-volume)
• Backport ceph/ceph@0358bd5 (another fix for ceph-volume)
• Handle GCC-14 compile errors
• Rebuild / fix all the fun problems I'm sure python 3.12 will bring
  • run-tox-mgr
  • run-tox-mgr-dashboard-py3
  • run-tox-mgr-dashboard-lint
  • run-tox-mgr-dashboard-openapi
  • run-tox-python-common (root error: Fake pathlib not working with Python 3.12-dev pytest-dev/pyfakefs#770, needs pyfakefs>=5.2.0)
  • run-tox-cephadm
  • run-tox-cephfs-top
  • unittest_mempool (looks like just a perf regression the upstream hasn't decided how to fix yet: test/test_mempool: skip check_shard_select test on ARM64 ceph/ceph#56448, mempool: Modify pick_a_shard_int() fucntion ceph/ceph#55249, common/mempool.cc: Improve performance of sharding ceph/ceph#55696)
  • test_ceph_argparse.py
  • unittest_deferred (disappeared after GCC-14 rebuild)

[bazaah]

The following tests FAILED:
          4 - run-tox-mgr (Failed)
          7 - run-tox-mgr-dashboard-py3 (Failed)
         10 - run-tox-mgr-dashboard-lint (Failed)
         16 - run-tox-mgr-dashboard-openapi (Failed)
         19 - run-tox-python-common (Failed)
         22 - run-tox-cephadm (Failed)
         28 - run-tox-cephfs-top (Failed)
         56 - unittest_mempool (Failed)
        148 - test_ceph_argparse.py (Failed)
        192 - unittest_deferred (Subprocess aborted)

unsurprisingly, all the python tests fail. More surprisingly some of the cpp ones do. Need to investigate what changed

[bazaah]

So I've finally had time to sit down and work on this. py3.12 is as annoying as I was expecting. I may end up pushing a pkgrel with parts of the python bits broken, but the rest of the project functional, we'll see.

[bazaah]

so we just got GCC 14, which has now exploded the build in other interesting ways. I'll need to investigate if the upstream has encountered these problems, because understanding C++ template expansion in macros is beyond my ability

[bazaah]

Pretty sure dashboard is broken again... I won't block the next pkgrel on fixing that

[bazaah]

I have fixed all the test failures excluding the stuff surrounding the around the dashboard (seemingly requests relies on a pyO3 powered bomb somewhere in its deptree). I'm doing a fresh rebuild now, and assuming nothing else breaks will do my integration testing and then push a new pkgrel

[bazaah]

What should be the final changeset is now being built. Should be pushed out in 5-6h

[bazaah]

done in 8d8f54d