| Age | Commit message (Collapse) | Author |
|---|
|
* HEAD
| \
| * 54c0f53 <github@luxolus.com> pkgbuild: pkgver=18.2.4 pkgrel=3
| * cc51155 <github@luxolus.com> pkgbuild: use fmt-formatter-const.patch
| * e717a88 <github@luxolus.com> pkgbuild: override Findfmt.cmake FMT_VERSION file lookup
| * 75a3706 <github@luxolus.com> repo: add fmt-formatter-const.patch
| /
| master
Signed-off-by: Bazaah <github@luxolus.com>
|
|
|
|
note this is a bit more complicated than the normal use-patch commit, as
part of the changes in fmt-formatter-const.patch were previously made in
fmt10-fixes.patch.
We remove those from fmt10-fixes, ensure the patch _we_ produce uses
const (...oops) and fixup file hashes
|
|
fmt11 has moved the definition of FMT_VERSION from core.h to base.h, so
we adjust the cmake function the upstream uses.
This isn't a patch because the next major version completely changes how
they look up fmt (and a lot of other things)
References: https://github.com/fmtlib/fmt/releases/tag/11.0.0
|
|
This patch ensures that _all_ fmt::formatter::format type functions
within the code base are const, as the fmt maintainers expressly mention
they should be.
Hopefully this fixes some of the random FTBS' we get when upgrading
boost/gcc/fmt
References: https://github.com/fmtlib/fmt/issues/2747
|
|
* HEAD
| \
| * 10719f1 <github@luxolus.com> pkgbuild: pkgver=18.2.4 pkgrel=2
| * 62c43be <github@luxolus.com> pkgbuild: use ceph-18.2.4-boost-1.86-fixes.patch
| * 8c5cc7d <github@luxolus.com> repo: add boost-1.86-fixes.patch
| /
| master
Signed-off-by: Bazaah <github@luxolus.com>
|
|
|
|
|
|
Fixes three issues found during rebuild:
1. Change to boost::uuids::uuid preventing *char cast
2. Missed header for boost::mt11213b
Seemingly an actual bug on ceph, this header should have been
included as far back as 1.60 but I guess was transitively included
from somewhere, and now isn't
3. undef BOOST_MPL_CFG_NO_PREPROCESSED_HEADERS before including
<boost/python/*> headers
Apparently, you win free debugging of boost metaprogramming during
preprocessor macro expansion if you do this; who knew?
The maintainers of boost python. Because of the open 2 year old bug
report. (:
References: https://github.com/boostorg/python/issues/394
References: https://www.boost.org/doc/libs/1_86_0/libs/uuid/doc/html/uuid.html#changes_changes_in_boost_1_86_0_major_update
References: https://www.boost.org/doc/libs/1_86_0/doc/html/doxygen/headers/mersenne__twister_8hpp_1a29ac4e21d84adda27beb1803a0525ca5.html
|
|
* HEAD
| \
| * 2423fcc <github@luxolus.com> pkgbuild: pkgver=18.2.4 pkgrel=1
| * b55971a <github@luxolus.com> pkgbuild: update license to use SPDX identifier
| * e432e29 <github@luxolus.com> pkgbuild: add ceph-exporter.service to pkg.ceph-mgr
| * 879e512 <github@luxolus.com> pkgbuild: add pkg.ceph-node-proxy
| * 90b4dfa <github@luxolus.com> pkgbuild: rm included backport (mgr-dashboard-simplify-authentication-protocol)
| * 5d846f6 <github@luxolus.com> pkgbuild: rm included backport (ceph-volume-check-generic-reject-reasons)
| * 335c26f <github@luxolus.com> pkgbuild: rm superseded patch (tox-mypy-false-postive)
| * be62f5b <github@luxolus.com> repo: rm zombie patch (delete-test-librados-asio)
| /
| master
Signed-off-by: Bazaah <github@luxolus.com>
|
|
|
|
Its somewhat unclear how exactly ceph is licensed overall, and I think
my choice of licenses is not *wrong*, but may be incomplete.
References: https://gitlab.archlinux.org/archlinux/rfcs/-/blob/master/rfcs/0016-spdx-license-identifiers.rst
References: https://github.com/ceph/ceph/blob/v18.2.4/COPYING-LGPL3
References: https://github.com/ceph/ceph/blob/v18.2.4/COPYING-LGPL2.1
References: https://github.com/ceph/ceph/blob/v18.2.4/COPYING-GPL2
|
|
New systemd unit file for the 'ceph-exporter' daemon. There is
seemingly no official documentation for this service but I think its an
inhouse prometheus exporter that runs per node... though unclear if
'node' means 'osd' or 'mon' or something else.
References: https://github.com/ceph/ceph/tree/v18.2.4/src/exporter
|
|
This is seemingly something to do with cephadm and (Dell?) hardware
metrics using the RedFishAPI... api. Apparently it was split out because
the upstream 'needed' it in v18 but the rest of the changes around it
from v19 weren't going to be backported.
Seems fairly worthless for ArchLinux but who can say.
References: https://github.com/ceph/ceph/commit/e1630fb8fcfa4460306b5e506c6a436f68c505a8
|
|
Upstream-ref: https://github.com/ceph/ceph/commit/33d8befcabe3e60d9ccf6e4fdc2598c4db22934b
|
|
Upstream-ref: https://github.com/ceph/ceph/commit/50cdc0b66ce87888f1e634ed4323d91bd7273dcb
|
|
Since 14aff06 we now completely disable mypy, so this patch is irrelevant
|
|
Missed in 4095603 when patch was removed from pkgbuild
|
|
* HEAD
| \
| * 91a08e5 <github@luxolus.com> pkgbuild: pkgver=18.2.2 pkgrel=4
| * c95ff54 <github@luxolus.com> pkgbuild: bump ceph_bcrypt to 4.1.3
| /
| master
Signed-off-by: Bazaah <github@luxolus.com>
|
|
|
|
|
|
* HEAD
| \
| * c994ac3 <github@luxolus.com> pkgbuild: pkgver=18.2.2 pkgrel=3
| * dea7754 <github@luxolus.com> pkgbuild: rm VERBOSE=1 output from make check
| * 5867453 <github@luxolus.com> pkgbuild: add python-packaging to makedeps, ceph-mgr
| * 40c2ca2 <github@luxolus.com> pkgbuild: add pkgrel=3 patches
| * b2d88d6 <github@luxolus.com> pkgbuild: use ceph-volume backport patches
| * 9e62fb3 <github@luxolus.com> pkgbuild: cp mgr's ceph_module mocks to a dummy module
| * 0fe6e33 <github@luxolus.com> pkgbuild: rm mgr-ceph-module-stub.patch
| * ffa09b1 <github@luxolus.com> pkgbuild: add ceph_bcrypt to test suites
| * 14aff06 <github@luxolus.com> repo: add disable-mypy-flake8-tests
| * 232f0f5 <github@luxolus.com> repo: add py312-fixes.patch
| * 3a2a24a <github@luxolus.com> repo: add test-mempool-shard-select-disable
| * 709961b <github@luxolus.com> repo: add std-atomic-depreciations.patch
| * d8b813a <github@luxolus.com> repo: add gcc-14-fixes
| * acc985a <github@luxolus.com> repo: backport two fixes for ceph-volume
| /
| master
Signed-off-by: Bazaah <github@luxolus.com>
|
|
|
|
|
|
|
|
|
|
|
|
Basically, we add the initialization for the mgr's test suite to a
location that *all* mgr module test suites include in their PYTHONPATH
I'm not sure why this worked before, when it stopped exactly, and what
changed to make it stop. Yay python.
This shouldn't be copied into the CMAKE_INSTALL_DIR, but we should
double check the built packages to be 100% sure.
|
|
This doesn't work. Or rather it does, but badly and for reasons I didn't
understand.
To set the stage.
1. All mgr modules in src/pybind/mgr/*/module.py import a 'mgr_module'
2. This module is located at src/pybind/mgr_module.py
3. It imports a module, 'ceph_module' that **does not exist.**
I didn't understand how this could work, as the file even remotely
similiar is src/pybind/mgr/ceph_module.pyi -- note the 'i'. This is what
is called a "python interface file" and basically isn't something the
python runtime is aware of. It's only use is by static code checkers
like mypy (and presumably the IDEs of people that work in this code).
I, naively assumed that somehow previous python versions had mistaken
this file for a real module hence the previous patch, where I just
define the various symbols referenced from other modules.
This isn't the case. Instead, the C++ mgr daemon _at runtime_ injects
what is effectively a bridge module into the C++ defined ceph data
structures.
This, obviously doesn't work during testing, but the upstream does have
some interesting machinery in src/pybind/mgr/tests/__init__.py, though I
don't understand how it would propagate to other test suites, and very
clearly it hasn't worked for at least since 18.2.0 for our builds.
A future commit will properly fix this issue
|
|
We use --file-links to hint to python that it should consider more
locations for package lookups.
Could be useful if python-xmlsec ever rears its ugly head again
|
|
A whole bunch of random lints and "x is best practice" messages have
started generating a bunch of unwanted line noise in the test suite, so
we disable them all
|
|
Fix a whole host of fun python issues:
- Use '<mock>.assert_called_with()' over 'assert <mock>.called_with()'
- Comment out failing TODO'd test for cephadm haproxy
- Use pyfakefs >=5.2 to fix a whole host of import errors
- Use packaging.version over pkg_resources.packaging.version
- Fix a stupid assert of a urlparse error message (exception messages
aren't stable, lads)
- Use assertRegex over assertRegexpMatches with unittest mocks
Also note that the test suite *still* fails even after these, as
seemingly the dashboard is completely broken by some PyO3 package, but
at least nothing there is expected to work
|
|
this is basically just a performance issue that is triggering on our
build server.
Should be reverted once the upstream decides how to fix it.
Quoting myself:
> looks like just a perf regression the upstream hasn't decided how
> to fix yet: ceph/ceph#56448, ceph/ceph#55249, ceph/ceph#55696
References: https://github.com/bazaah/aur-ceph/issues/23
|
|
Wrap a few instances of std::shared_ptr that should be protected by
std::atomic as of C++20
References: https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2023/p2869r0.pdf
|
|
Backports a few fixes for GCC 14 that some of the upstream's distro
builds have encountered.
This takes from the SUSE's leap stream, though we had to cut out parts,
so it isn't a "real" backport.
References: https://github.com/SUSE/ceph/pull/516
|
|
Unclear why at least the former hasn't been included in a .3 release,
yet as it actually prevents the command from executing.
Regardless, we do it here.
Upstream-Ref: https://github.com/ceph/ceph/commit/9ee47b03286ce5a8de167abbd592db7b5aedc4fc
Upstream-Ref: https://github.com/ceph/ceph/commit/0358bd5ea27a7a6487f71cfb15b6bab536d04f74
|
|
* HEAD
| \
| * c8a69c3 <github@luxolus.com> pkgbuild: pkgrel=2
| * a6b3328 <github@luxolus.com> pkgbuild: use mgr-alias-ceph-bcrypt.patch
| * 8f49d39 <github@luxolus.com> repo: add mgr-alias-ceph-bcrypt.patch
| /
| master
Signed-off-by: Bazaah <github@luxolus.com>
|
|
|
|
|
|
So that the mgr _actually_ uses our internal bcrypt fork.
|
|
* HEAD
| \
| * c719c85 <github@luxolus.com> pkgbuild: pkgver=18.2.2 pkgrel=1
| * 3d35081 <github@luxolus.com> pkgbuild: add ceph_bcrypt to python-ceph-common
| * 8b874b8 <github@luxolus.com> pkgbuild: add python-bcrypt sources,patches
| * 8576de3 <github@luxolus.com> repo: add python-bcrypt patches
| * 246af22 <github@luxolus.com> pkgbuild: restrict prepare() patch machinery to ceph-*
| * fcde70e <github@luxolus.com> pkgbuild: mv {,ceph-}disable-empty-readable.sh-test
| * abc0a56 <github@luxolus.com> pkgbuild: use mgr-ceph-module-stub.patch
| * a75f61e <github@luxolus.com> repo: add ceph-18.2.2-mgr-ceph-module-stub.patch
| * 93719e4 <github@luxolus.com> pkgbuild: use mgr-dashboard fix backport
| * 836304d <github@luxolus.com> repo: backport mgr/dashboard fix for python-cryptography issues
| * bfe55db <github@luxolus.com> pkgbuild: rm checkdepends->python-mock
| /
| master
Signed-off-by: Bazaah <github@luxolus.com>
|
|
|
|
|
|
For our internal ceph fork of python-bcrypt
|
|
- python-bcrypt-allow-subinterpreters.patch
- python-bcrypt-prefix-ceph.patch
Together, these allow us to build a renamed python-bcrypt package, as
ceph_bcrypt.
This will allow us to bypass the thorny pyo3 issues we have with various
mgr modules exploding at runtime.
References: https://github.com/bazaah/aur-ceph/issues/20
|
|
So that we can introduce non ceph patches in future commits
|
|
|
|
|
|
This mocks out a module for mgr_module.py to 'import' during mypy
type analysis. Previously the ceph_module.pyi was enough for this to
work, but I guess something in the tooling changed.
I cannot figure out why this ever worked before, and why it doesn't
now.
Maybe related to https://github.com/python/mypy/issues/16987 ?
|
|
and remove the now unneeded py deps:
- python-pyjwt
- python-cryptography
|