Kefu Chai [Wed, 28 Jan 2026 02:58:31 +0000 (10:58 +0800)]
pybind/rbd: move legacy_implicit_noexcept to rbd.pyx
Move the legacy_implicit_noexcept compiler directive from setup.py to
the top of rbd.pyx, making it consistent with how CephFS handles this
directive. This simplifies the build setup by:
- Removing conditional logic based on Cython version in setup.py
- Eliminating the need for compiler_directives dict and packaging import
- Making RBD's directive handling consistent with other bindings
The directive is needed for building with both Cython 0.x and Cython 3
from the same file while preserving the same behavior. Cython safely
ignores unknown compiler directives when specified at the top of .pyx
files, so this works across all supported Cython versions.
When Cython 0.x support is eventually dropped, this directive can be
replaced with explicit noexcept annotations on rbd_callback_t and
librbd_progress_fn_t type definitions.
Kefu Chai [Fri, 23 Jan 2026 01:36:22 +0000 (09:36 +0800)]
pybind: hardwire language_level to 3
Previously, to maintain backward compatibility with Python 2, we set
'language_level' to sys.version_info.major, so the value would be 2
when building with Python 2, and 3 with Python 3. Now that Python 2
support has been dropped, we can hardwire it to "3".
This change also removes the comment about switching to
`language_level=3str` in the future. According to the Cython 3.1+
documentation,
> language_level=3 is now the default. language_level=3str has become a
> legacy alias.
see https://cython.readthedocs.io/en/3.1.x/src/changes.html.
For context, in Cython < 3.1, language_level=3 and language_level=3str
had different meanings:
- 3 = unprefixed strings are unicode
- 3str = unprefixed strings follow Python version (bytes in Py2, unicode
in Py3)
Since we no longer support Python 2, this distinction is irrelevant and
the comment can be safely removed.
Kefu Chai [Tue, 27 Jan 2026 07:08:28 +0000 (15:08 +0800)]
cmake: migrate Python module installation from setup.py to pip
Replace 'setup.py install' with 'pip install --use-pep517' to fix
Cython compilation failures and eliminate deprecation warnings.
Problem Statement:
The build process for Cython modules involves preprocessing .pyx files
(e.g., generating rbd_processed.pyx from rbd.pyx) and then cythonizing
with specific compiler_directives. The previous approach using separate
'setup.py build' and 'setup.py install' commands caused this failure:
rbd_processed.pyx:781:44: Cannot assign type 'int (*)(uint64_t, uint64_t, void *) except? -1' to 'librbd_progress_fn_t'. Exception values are incompatible. Suggest adding 'noexcept' to type 'int (uint64_t, uint64_t, void *) except? -1'.
```
This occurs because:
1. 'setup.py build build_ext' successfully preprocesses and cythonizes
with compiler_directives from setup.py's cythonize() call
2. 'setup.py install' internally triggers a rebuild that:
- Regenerates the preprocessed .pyx files
- Re-runs cythonize() through Cython.Distutils.build_ext
- Does NOT apply the compiler_directives from setup.py
- Fails on the regenerated files missing required directives
New Options Explained:
`--use-pep517`:
Addresses deprecation warning:
```
DEPRECATION: Building 'rados' using the legacy setup.py bdist_wheel
mechanism, which will be removed in a future version. pip 25.3 will
enforce this behaviour change.
```
Uses the modern PEP 517 build backend which:
- Performs a single build pass with all compiler_directives applied
- Prevents the implicit rebuild that caused CompileError
- Future-proofs against pip 25.3+ which will require this
`--no-build-isolation`:
Ensures that environment variables set by CMake are respected:
- CC, LDSHARED (compiler toolchain)
- CPPFLAGS, LDFLAGS (compilation flags)
- CYTHON_BUILD_DIR, CEPH_LIBDIR (build paths)
Without this flag, pip would create an isolated build environment
that ignores these critical build settings.
`--no-deps`:
Prevents pip from attempting to install Python dependencies listed
in setup.py's install_requires. All dependencies are managed by
CMake and the distribution's package manager, not pip.
`--ignore-installed`:
Addresses installation error when DESTDIR is set:
```
ERROR: Could not install packages due to an OSError: [Errno 13]
Permission denied: '/usr/lib/python3/dist-packages/rados-2.0.0.egg-info'
OSError: [Errno 18] Invalid cross-device link:
'/usr/lib/python3/dist-packages/rados-2.0.0.egg-info' -> '/tmp/pip-uninstall-...'
```
This error occurs because pip detects an existing system installation
and tries to uninstall it before installing to DESTDIR. With
--ignore-installed, pip skips the uninstall step and directly installs
to the DESTDIR staging directory, which is the correct behavior for
packaging.
Removed Options:
`--install-layout=deb`:
This Debian-specific patch to 'setup.py install' is no longer needed.
Modern pip automatically detects the distribution and uses the correct
layout (dist-packages on Debian, site-packages on RPM distros).
`--single-version-externally-managed`:
This option was specific to 'setup.py install' to prevent egg
installation. With pip, this is handled automatically.
`--record /dev/null`:
No longer needed as pip manages installation records internally.
`egg_info --egg-base`:
Not needed with pip as metadata is generated automatically during
the build process.
Not added option:
`--root-user-action=ignore`: not added
In this change, we installing a python module using pip with
`fakeroot` before packaging it. But pip warned:
```
Error: WARNING: Running pip as the 'root' user can result in broken permissions and conflicting behavior with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv
```
But we use fakeroot on purpose, this option could have been added to
silence this warning. But it is not available in all supported pip
versions. see
https://github.com/pypa/pip/commit/2e1112a8141dbdf767505fded918706e9ad61031
New environmental variable:
`DEB_PYTHON_INSTALL_LAYOUT=deb` is conditionally applied when packaging
for debian-derivative distributions. As pip does not support
`--install-layout` option. Since debian patches pip so it installs Python
modules into /usr/local/lib instead of /usr/lib where debian dh_install
helper looks for the content to be packaged, so we have to enforce the
debian layout using the environmental variable.
Working Directory Change:
Changed from `CMAKE_CURRENT_SOURCE_DIR` to `CMAKE_CURRENT_BINARY_DIR` to
keep pip's temporary files and logs in the build directory rather than
polluting the source tree.
Additional Dependencies:
Since the build process uses pip and creates a wheel distribution,
we need to add `pip` and `wheel` Python modules as build dependencies.
Python moduels packaging:
- with `--use-pep517`, pip creates .dist-info directoires as per PEP-517
instead of .egg-info, so we need to package the new metadata directory.
Future Improvements
We considered implementing a custom `build_templates` command or using
setuptools' `sub_commands` mechanism to avoid regenerating `*_processed.pyx`
files on every build (tracking dependencies via file modification times or
hash-based checks). However, to keep `setup.py` simple and maintainable,
we've deferred this optimization for future work. The current solution
using `pip install --use-pep517` ensures correct builds without additional
complexity.
This solution works correctly for both Debian and RPM packaging workflows,
both of which use DESTDIR-based staged installations.
Ilya Dryomov [Wed, 28 Jan 2026 09:41:13 +0000 (10:41 +0100)]
qa/workunits/rbd: drop randomized sleeps in "big image" tests
These tests were tuned for slower hardware than what we have now.
Even without these the image is often 25-30% synced by the time the
test gets to the "non-primary snapshot in question is still being
synced" assert.
Ilya Dryomov [Tue, 27 Jan 2026 20:56:23 +0000 (21:56 +0100)]
qa/workunits/rbd: avoid unnecessary sleeping in stop_mirror()
There is no need to wait for anything if -KILL is passed for sig
because the process would disappear immediately. In teuthology runs
where multiple rbd-mirror daemons are deployed (and therefore need to
be stopped when stop_mirrors() is called by the test), it causes
gratuitous delays of 4+ seconds.
Kefu Chai [Tue, 27 Jan 2026 08:34:09 +0000 (16:34 +0800)]
test/ceph_assert: fix death test timeout with threadsafe style
The unittest_ceph_assert test was timing out intermittently with warning:
```
"Death tests use fork(), which is unsafe particularly in a threaded
context. For this test, Google Test detected 3 threads."
```
The issue occurs because global_init() creates threads before the death
tests run. Use the "threadsafe" death test style to handle this safely,
matching the pattern already used in ceph_breakpad.cc and
test_perf_counters_cache.cc.
see also https://github.com/google/googletest/blob/main/docs/advanced.md#death-tests-and-threads
mgr/dashboard:Notification Footer+ Notification Page Fixes: https://tracker.ceph.com/issues/71738
A comprehensive notifications management page that allows users to view, search, and manage system notifications with a modern Carbon Design System interface.
Ilya Dryomov [Fri, 23 Jan 2026 13:48:53 +0000 (14:48 +0100)]
qa: don't assume that /dev/sda or /dev/vda is present in unmap.t
Instead of hard-coding the block device name, use the block device that
is backing the filesystem that the test is running on. We can be quite
sure it won't be an RBD device ;)
Imran Imtiaz [Thu, 8 Jan 2026 10:37:32 +0000 (10:37 +0000)]
mgr/dashboard: fix RBD mirror schedule inheritance in pool and image APIs
Signed-off-by: Imran Imtiaz <imran.imtiaz@uk.ibm.com> Fixes: https://tracker.ceph.com/issues/74494
Fix the bug where the Pool API was reporting random image schedules
instead of pool schedules. Implement proper schedule inheritance
hierarchy (Image > Pool > Cluster) for both Pool and Image APIs.
Disable OSD bench from benchmarking the OSDs for teuthology tests. This is to
help prevent a cluster warning pertaining to the IOPS value not lying within
a typical threshold range from being raised.
The tests can rely on the built-in static values as defined by
osd_mclock_max_capacity_iops_[ssd|hdd] which should be good enough.
This should have been a default in teuthology.yaml. My reasoning for the revert is:
- This is hardcoding a value
- Folks wouldn't know this is configurable without looking at cephadm.py
- Instead of doing what the error message said to do, we removed the error message
Signed-off-by: David Galloway <david.galloway@ibm.com>
Ilya Dryomov [Wed, 21 Jan 2026 18:41:41 +0000 (19:41 +0100)]
qa: krbd_blkroset.t: eliminate a race in the open_count test
Even at QD=1, dd may take less than 10 seconds to work its way to the
end of a 10M image, producing "No space left on device" error instead
of the expected "Operation not permitted" error which is supposed to
arise from the device getting marked read-only while opened.
With https://github.com/ceph/ceph-build/pull/2497 merged we no loger
build Tentacle+Crimson regularly. As Crimson no longer backport changes
into Tentacle, there's no reason to keep testing it.
Matan Breizman [Thu, 22 Jan 2026 10:00:25 +0000 (12:00 +0200)]
container/build.sh: Use dedicated debug tags
https://github.com/ceph/ceph-build/pull/2497 introduced a debug flavor.
This seems to cause conflicts with the image being pushed to quay as one
of the flavors might override the other.
Tag debug build containers explicitly.
Alternative solution would be to skip debug containers all together.
However. these might be useful for development purposes.
Note, prune-quay might also need to be updated once this is merged.
Kefu Chai [Thu, 22 Jan 2026 03:57:37 +0000 (11:57 +0800)]
cmake: fix undefined PY_LDFLAGS in distutils_install_cython_module
The distutils_install_cython_module() function was using ${PY_LDFLAGS}
without defining it, causing the linker to fail with:
/opt/rh/gcc-toolset-13/root/usr/libexec/gcc/x86_64-redhat-linux/13/ld:
cannot find -lrados: No such file or directory
This bug was introduced in commit d22734f6cb0 which changed:
set(ENV{LDFLAGS} "-L${CMAKE_LIBRARY_OUTPUT_DIRECTORY}")
to:
set(ENV{LDFLAGS} "${PY_LDFLAGS}")
However, PY_LDFLAGS was only defined in distutils_add_cython_module(),
not in distutils_install_cython_module(). This meant that during the
install phase, LDFLAGS was set to an empty string, and the linker
couldn't find librados.so and other Ceph libraries in the build
directory.
The bug was exposed by commit 719b74984605b490f23004eb41583a22c934c5fb
which changed rados.pxd to use C preprocessor conditionals (#ifdef
BUILD_DOC) instead of Cython's compile-time IF statements. This meant
the build now required proper linking during the install phase.
Fix by defining PY_LDFLAGS in distutils_install_cython_module():
Ville Ojamo [Fri, 16 Jan 2026 09:43:31 +0000 (16:43 +0700)]
doc/radosgw: change all intra-docs links to use ref (2 of 6)
Part 2 of 6 to make backporting easier. Depends on part 1.
Use the the ref role for all remaining links in doc/radosgw/ with the
exception of config-ref.rst which will depend on changes to rgw.yaml.in.
The external link definitions syntax being removed is intended for
linking to external websites and not for intra-docs links. Validity of
ref links will be checked during the docs build process.
Add labels for links targets if necessary.
Remove unused external link definitions in the modified files.
Use confval instead of literal text for 2 configuration keys in
vault.rst.
Signed-off-by: Ville Ojamo <14869000+bluikko@users.noreply.github.com>
Ville Ojamo [Fri, 16 Jan 2026 08:55:27 +0000 (15:55 +0700)]
doc/radosgw: change all intra-docs links to use ref (1 of 6)
Part 1 of 6 to make backporting easier. Many of the following parts
depend on this.
Use the the ref role for all remaining links in doc/radosgw/ with the
exception of config-ref.rst which will depend on changes to rgw.yaml.in.
The external link definitions syntax being removed is intended for
linking to external websites and not for intra-docs links. Validity of
ref links will be checked during the docs build process.
Add labels for links targets if necessary.
Remove unused external link definitions in the modified files.
Use confval instead of literal text for 2 configuration keys in
vault.rst.
Use Ceph Object Gateway consistently in multisite.rst.
Signed-off-by: Ville Ojamo <14869000+bluikko@users.noreply.github.com>
Ronen Friedman [Wed, 21 Jan 2026 12:37:24 +0000 (14:37 +0200)]
Merge pull request #66626 from ronen-fr/wip-rf-aborthp-justdoc
doc/ceph.rst: scrub-related 'tell pgid' commands
Related to https://github.com/ceph/ceph/pull/66515 Reviewed-by: Anthony D'Atri <anthony.datri@gmail.com> Reviewed-by: Ville Ojamo <14869000+bluikko@users.noreply.github.com>
Afreen Misbah [Tue, 13 Jan 2026 20:47:40 +0000 (02:17 +0530)]
mgr/dashboard: Add productive card component
- add generic productive card component
- based on carbon design system
- there are two versions of card - with shadow(tinted affect) and without.
- applies gray10 theme which is decided by new designs.
Fix on_operator_abort_scrub() to handle the case where
the operator-initiated abort request arrives while the
'start scrub' message is still in the queue (i.e. -
is_queued_or_active() is true, but is_scrub_active()
is false).
Unlike our handling of, for example, FullReset in
PrimaryIdle::clear_state(), here we choose to ignore
the request:
Considering the added complexity to the FSM versus
the minimal benefit, it is better to just ignore this
very rare case, leaving it to the operator to re-issue
the abort command if needed.
Ronen Friedman [Thu, 4 Dec 2025 14:49:29 +0000 (08:49 -0600)]
osd/scrub: support an operator-abort command
The new explicit command aborts any ongoing scrub of the target PG,
including operator-initiated scrubs. That additional capability is needed now that
operator-initiated scrubs are no longer blocked by 'no-scrub' settings.
The scenario we are trying to help the operator with is:
- an operator issues a set of operator-initiated scrubs (e.g., via a
script), then realizes the mistake and wants to abort them all.
The abort command also downgrades the urgency level of the scrub target
(as otherwise the target would immediately restart, against the operator
wishes).
This commit implements the changes to the state machine and to the abort
logic, assuming the operator command was translated into an event.
Ville Ojamo [Tue, 20 Jan 2026 06:17:44 +0000 (13:17 +0700)]
doc/rados: fix links in operations/cache-tiering.rst
Change a link using an external link definition to ref that was missed
in commit 49af82c, PR #66943.
Remove a sentence that also linked to the same destination because the
destination had no cache tier configuration or default values that the
text promised.
Signed-off-by: Ville Ojamo <14869000+bluikko@users.noreply.github.com>
Gil Bregman [Mon, 19 Jan 2026 12:18:03 +0000 (14:18 +0200)]
mgr/cephadm: Add some new fields to the cephadm NVMEoF spec file. Fixes: https://tracker.ceph.com/issues/74446 Signed-off-by: Gil Bregman <gbregman@il.ibm.com>
Aashish Sharma [Wed, 17 Dec 2025 09:21:14 +0000 (14:51 +0530)]
monitoring: make cluster matcher backward compatible for pre-7.1 metrics
Ceph 18.* adds a `cluster` label to all Prometheus metrics. When
upgrading from earlier releases, historical metrics lack this label
and are excluded by Grafana queries that strictly match on `cluster`.
Update the shared Grafana matcher logic to use a regex matcher that
also matches series without the `cluster` label, restoring visibility
of pre-upgrade metrics while preserving multi-cluster behavior.
projects / ceph.git / log