Ilya Dryomov [Fri, 12 Aug 2022 09:10:45 +0000 (11:10 +0200)]
rbd: remove incorrect use of std::includes()
- std::includes() requires sorted ranges but command specs aren't
sorted
- std::includes() purpose is to check whether the second range is
a subsequence of the first range but here the size of the second
range is always equal to the size of the first range, which means
that, had the ranges been sorted, std::includes() would have checked
straight equality
Ilya Dryomov [Fri, 12 Aug 2022 09:10:45 +0000 (11:10 +0200)]
rbd: find_action() should sort actions first
The order in which objects with static storage duration in
different TUs are initialized is undefined. If the compiler
chooses to initialize Shell::Action objects in action/Trash.cc
before Shell::Action objects in action/TrashPurgeSchedule.cc,
all "rbd trash purge schedule ..." commands get shadowed by
"rbd trash purge" command:
$ rbd trash purge schedule list
rbd: too many arguments
The confusing error arises because "rbd trash purge" takes a single
positional argument. "schedule" gets interpreted as <pool-spec> and
"list" generates an error.
Kefu Chai [Wed, 10 Aug 2022 14:35:14 +0000 (22:35 +0800)]
ceph.spec.in: %enable_devtoolset11 only if the macro is defined
there is chance that we are using `yum-builddep` to prepare the
build dependencies. in that case, gcc-toolset-11-build is not
installed. it's like a chicken-egg dilemma, but the point is
`yum-builddep` is able to pull in the gcc-toolset-11-build. once
gcc-toolset-11-build is installed, we will have the %enable_devtoolset11
rpm macro.
Kefu Chai [Mon, 8 Aug 2022 12:40:52 +0000 (20:40 +0800)]
ceph.spec.in: add libatomic to BuildRequires on fedora
otherwise we'd have failures like
/opt/compiler-explorer/gcc-trunk-20220808/bin/../lib/gcc/x86_64-linux-gnu/13.0.0/../../../../x86_64-linux-gnu/bin/ld:
/tmp/ccVlMbVh.o: in function `std::atomic<tagged_ptr>::store(tagged_ptr,
std::memory_order)':
/opt/compiler-explorer/gcc-trunk-20220808/include/c++/13.0.0/atomic:273:
undefined reference to `__atomic_store_16'
when generating the building system using CMake on fedora 36.
Kefu Chai [Thu, 4 Aug 2022 03:57:05 +0000 (11:57 +0800)]
ceph.spec.in: BuildRequires gcc-toolset-11-libatomic-devel for aarch64
to address following failure when generating the building system
using CMake:
```
-- Performing Test HAVE_LIBATOMIC
-- Performing Test HAVE_LIBATOMIC - Failed
CMake Error at cmake/modules/CheckCxxAtomic.cmake:66 (message):
Host compiler /opt/rh/gcc-toolset-11/root/usr/bin/c++ requires libatomic,
but it is not found
```
Kefu Chai [Thu, 4 Aug 2022 01:00:29 +0000 (09:00 +0800)]
ceph.spec.in: call %enable_devtoolset11 on centos8
before this change %enable_devtoolset11 is called only when building
with crimson on centos8.
after this change %enable_devtoolset11 is called when building on
centos8. because we've started using gcc-toolset-11 for building
rpm packages on centos8 after the C++20 migration. so, to build
with gcc-11, we need to enable it.
also, because gcc-toolset-11 is used, we have to disable
annotated_build.
gcc-toolset-11-annobin is already installed, but ceph.spec.in adds
"-specs=/usr/lib/rpm/redhat/redhat-annobin-cc1" which needs the gcc
plugin too
resolves this failure during the cmake configure step:
-- Check for working CXX compiler: /opt/rh/gcc-toolset-11/root/usr/bin/c++
-- Check for working CXX compiler: /opt/rh/gcc-toolset-11/root/usr/bin/c++ - broken
CMake Error at /usr/share/cmake/Modules/CMakeTestCXXCompiler.cmake:59 (message):
The C++ compiler
Kefu Chai [Mon, 27 Jun 2022 13:18:57 +0000 (21:18 +0800)]
ceph.spec.in: use %enable_devtoolset11 to enable GTS-11
%enable_devtoolset11 redefines %___build_pre by appending
`source scl_source enable gcc-toolset-11` to it. `___build_pre` should
be able to populate this setting to both %build and %install. and hence
address the FTBFS where we need to use the tool chain from GTS-11.
This can be surprising but we actually compile things during
the `install` stage of `rpm-build`. The example is the pybind's
`setup.py` which builds `rados_dummy.c`.
ceph.spec.in: use gcc-toolset-10 for building crimson
This commit bumps up the toolset version but only to build crimson.
That is, the classical OSD stays unaffected.
The reason behind the upgrade is the following FTBFS:
```
[ 32%] Building CXX object src/seastar/CMakeFiles/seastar.dir/src/core/reactor.cc.o
/home/jenkins-build/build/workspace/ceph-dev-build/ARCH/x86_64/AVAILABLE_ARCH/x86_64/AVAILABLE_DIST/centos8/DIST/centos8/MACHINE_SIZE/gigantic/release/17.0.0-11345-ga3bb1485/rpm/el8/BUILD/ceph-17.0.0-11345-ga3bb1485/src/seastar/src/core/reactor.cc: In constructor ‘seastar::reactor::reactor(std::shared_ptr<seastar::smp>, seastar::alien::instance&, unsigned int, seastar::reactor_backend_selector, seastar::reactor_config)’:
/home/jenkins-build/build/workspace/ceph-dev-build/ARCH/x86_64/AVAILABLE_ARCH/x86_64/AVAILABLE_DIST/centos8/DIST/centos8/MACHINE_SIZE/gigantic/release/17.0.0-11345-ga3bb1485/rpm/el8/BUILD/ceph-17.0.0-11345-ga3bb1485/src/seastar/src/core/reactor.cc:926:90: error: use of deleted function ‘seastar::condition_variable::condition_variable()’
926 | , _thread_pool(std::make_unique<thread_pool>(this, seastar::format("syscall-{}", id))) {
| ^
In file included from /home/jenkins-build/build/workspace/ceph-dev-build/ARCH/x86_64/AVAILABLE_ARCH/x86_64/AVAILABLE_DIST/centos8/DIST/centos8/MACHINE_SIZE/gigantic/release/17.0.0-11345-ga3bb1485/rpm/el8/BUILD/ceph-17.0.0-11345-ga3bb1485/src/seastar/include/seastar/core/reactor.hh:74,
from /home/jenkins-build/build/workspace/ceph-dev-build/ARCH/x86_64/AVAILABLE_ARCH/x86_64/AVAILABLE_DIST/centos8/DIST/centos8/MACHINE_SIZE/gigantic/release/17.0.0-11345-ga3bb1485/rpm/el8/BUILD/ceph-17.0.0-11345-ga3bb1485/src/seastar/src/core/reactor.cc:32:
/home/jenkins-build/build/workspace/ceph-dev-build/ARCH/x86_64/AVAILABLE_ARCH/x86_64/AVAILABLE_DIST/centos8/DIST/centos8/MACHINE_SIZE/gigantic/release/17.0.0-11345-ga3bb1485/rpm/el8/BUILD/ceph-17.0.0-11345-ga3bb1485/src/seastar/include/seastar/core/condition-variable.hh:157:5: note: ‘seastar::condition_variable::condition_variable() noexcept’ is implicitly deleted because its exception-specification does not match the implicit exception-specification ‘’
157 | condition_variable() noexcept = default;
```
osd: Handle oncommits and wait for future work items from mClock queue
When a worker thread with the smallest thread index waits for future work
items from the mClock queue, oncommit callbacks are called. But after the
callback, the thread has to continue waiting instead of returning back to
the ShardedThreadPool::shardedthreadpool_worker() loop. Returning results
in the threads with the smallest index across all shards to busy loop
causing very high CPU utilization.
The fix involves reacquiring the shard_lock and waiting on sdata_cond
until notified or until time period lapses. After this, the smallest
thread index repopulates the oncommit queue from the context_queue
if there were any additions.
Kefu Chai [Fri, 5 Aug 2022 08:40:41 +0000 (16:40 +0800)]
cmake: remove spaces in macro used for compiling cython code
we are facing following FTBFS on jammy + GCC-11.2 + Cython 0.29 +
CMake 3.22:
creating /home/jenkins-build/build/workspace/ceph-api/build/lib/cython_modules/temp.linux-x86_64-3.10/home/jenkins-build/build/workspace/ceph-api/build/src/pybind/cephfs
compile options: '-I/usr/include/python3.10 -I/usr/include/python3.10 -c'
extra options: '-Wno-unused-result -Wsign-compare -DNDEBUG -g -fwrapv -O2 -Wall -g -fstack-protector-strong -Wformat -Werror=format-security -g -fwrapv -O2 -iquote/home/jenkins-build/build/workspace/ceph-api/src/include -w -Dvoid0=dead_function(void) -D__Pyx_check_single_interpreter(ARG)=ARG ## 0 -Wno-unused-result -Wsign-compare -DNDEBUG -g -fwrapv -O2 -Wall -g -fstack-protector-strong -Wformat -Werror=format-security -g -fwrapv -O2'
cc: /home/jenkins-build/build/workspace/ceph-api/build/src/pybind/cephfs/cephfs.c
cc: warning: ##: linker input file unused because linking not done
cc: error: ##: linker input file not found: No such file or directory
cc: warning: 0: linker input file unused because linking not done
cc: error: 0: linker input file not found: No such file or directory
it seems cython is not able to escape the space in the "extra options"
anymore, so the "##" and "0" are considered as object files passed to
compiler in addition to cephfs.c.
in this change the spaces are removed to help cython to make the right
decision.
We don't backport PRs merged into doc/releases. Therefore, when one browses to an older Ceph release version on docs.ceph.com (e.g., https://docs.ceph.com/en/pacific/), the information is out of date at best.
The doc/releases page is only accurate if browsing https://docs.ceph.com/en/latest/, for example.
So this post_checkout command will make sure we've checked out doc/releases from main before building and publishing.
Luis Domingues [Tue, 19 Jul 2022 09:04:34 +0000 (11:04 +0200)]
mgr/cephadm: add parsing for config on osd specs
Cephadm, while parsing spec files, can parse ceph configuration
for almost all the services, except for OSDs, where it fails
with a nasty "unexpected keyword argument config".
mgr/volumes: Fix subvolume creation in FIPS enabled system.
The md5 checksum is used in the construction of legacy
subvolume config filename. It's not used for security reason.
Hence marking the 'usedforsecurity' flag to false to
make it FIPs compliant.
The usage of md5 was always in there. The commit 373a04cf734
made it to get exercised in 'open_subvol' which is pre-requisite
for all the subvolume operations and hence subvolume
creation has failed.
When the class `Device` is instantiated with a path instead of a
block device, it fails like following.
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/usr/lib/python3.6/site-packages/ceph_volume/util/device.py", line 130, in __init__
self._parse()
File "/usr/lib/python3.6/site-packages/ceph_volume/util/device.py", line 233, in _parse
self.ceph_device = disk.has_bluestore_label(self.path)
File "/usr/lib/python3.6/site-packages/ceph_volume/util/disk.py", line 906, in has_bluestore_label
with open(device_path, "rb") as fd:
IsADirectoryError: [Errno 21] Is a directory: '/var/lib/ceph/osd/ceph-0/'
```
passing a path instead of a block device is valid, `simple scan` needs it.
ceph-volume unit tests shouldn't actually create contents on the
filesystem from where it runs (even though they are written in a tmp
dir), let's use pyfakefs.
ceph-volume shouldn't report devices `/dev/sdy` and `/dev/sdz`, they will never be
available in such a scenario. Considering this, in a host with a bunch of mpath devices
it will pollute the inventory output.
Zack Cerza [Tue, 21 Jun 2022 17:28:30 +0000 (11:28 -0600)]
ceph-volume: Rename env var; add warning
So that we can have a nice big warning that fires once per invocation, I
think using a callable class with a class attribute seems like a decent
approach. A closure could work too.
Zack Cerza [Tue, 17 May 2022 17:29:02 +0000 (11:29 -0600)]
ceph-volume: Optionally consume loop devices
A similar proposal was rejected in #24765; I understand the logic
behind the rejection, but this will allow us to run Ceph clusters on
machines that lack disk resources for testing purposes. We just need to
make it impossible to accidentally enable, and make it clear it is
unsupported.
This check was added in commit ecd3778a6f9a ("rbd-mirror: ensure that
the last non-primary snapshot cannot be pruned") as an additional
safeguard against pruning an incomplete non-primary snapshot in case
there is no predecessor mirror snapshot. However it still fires if the
predecessor is there but happens to be a primary demotion snapshot.
A bogus "incomplete local non-primary snapshot" error is reported and
the replayer gets stuck.
Remove completed_non_primary_snapshots_exist tracking as the presence
of the predecessor in the incomplete non-primary snapshot pruning arm
is already ensured by "m_local_snap_id_start > 0" condition.
Incomplete non-primary snapshot handling is bifurcated depending
on whether any data objects have been copied. If no data objects
have been copied, an incomplete non-primary snapshot is assumed to
be malformed and gets pruned; the sync is restarted from scratch.
It doesn't really thrash anything, just repeatedly restarts the
workload on top of a dirty cache file. rbd_pwl_cache_recovery is
more on point and gets covered by existing CODEOWNERS.
Yin Congmin [Fri, 7 Jan 2022 07:03:44 +0000 (15:03 +0800)]
qa/tasks: add thrash test for persistent write log cache
add thrash test for persistent write log cache. run rbd bench
on persistent write log cache, thrashes rbd bench, test the
recovery function of persistent write log cache.
projects / ceph.git / log