add a static library named global-static, which does not link with
libceph-common. so the executables which does not link against
lib{rados,cephfs,rbd} can be linked against global-static instead if
they want to access the symbols previously available from libglobal.
and libglobal is now linked against libceph-common. and it is supposed
to be used by executables packaged by ceph-test. these exectuables can
safely depend on libceph-common offered by package of "librados2".
Signed-off-by: Kefu Chai <kchai@redhat.com>
when packaging debian packages, dpkg-shlibdeps analyzes the linked
shared libraries of the binaries in given package by looking at their
names. but if the name does not include any useful versioning
information, it complains. we have no intention of versioning
libceph-common, as it is merely an internal shared library used by ceph
packages only. but there is no simple way to disable dpkg-shlibdeps'
warning of
dpkg-shlibdeps: warning: can't extract name and version from library
name 'libceph-common.so'
other than skipping the dpkg-shlibdeps for the whole package. so let's
just version it anyway.
Signed-off-by: Kefu Chai <kchai@redhat.com>
An allocation of size 0x80000000 gets truncated to 0. Larger values can
break things in similar ways.
Introduced in 5ab034345d.
Fixes: http://tracker.ceph.com/issues/18595
Signed-off-by: Sage Weil <sage@redhat.com>
Note that the MDS callers have new #warnings indicating that they
are not providing the pmore argument and are thus broken. (They
were already broken.)
Signed-off-by: Sage Weil <sage@redhat.com>
If we truncate the results of the omap read commands,
provide a flag so that the caller knows there is more
to be read. We don't need to provide the name of the
next key because the interface is defined as
"start_after" (not "start_with"), allowing them to use
the last key they received.
Signed-off-by: Sage Weil <sage@redhat.com>
Currently it only allows you to move buckets, which is annoying and much
less useful. To move an OSD you need to use create-or-move, which is
harder to use.
Fixes: http://tracker.ceph.com/issues/18587
Signed-off-by: Sage Weil <sage@redhat.com>
the dirfrag can be migareted to other mds while waiting in the
timer.
Fixes: http://tracker.ceph.com/issues/18487
Signed-off-by: "Yan, Zheng" <zyan@redhat.com>
This commit clarifies that folks opening bugfix PRs that need backporting to
one or more stable branches should always target master, and that the
backporting itself takes place only after the master PR is merged.
Signed-off-by: Nathan Cutler <ncutler@suse.com>
Before this patch the flag was wrongly handled in the Swift API
implementation. In rare conditions this might result in setting
req_state::system_request.
This may happen only if both of those conditions are fulfilled:
* RadosGW is running in a multi-site configuration (at least
one user with the system flag turned on is present),
* the "rgw_swift_account_in_url" configurable has been switched
to true. The value is false by default and our documentation
doesn't actually mention about the option.
The issue doesn't affect Jewel nor any previous release.
Fixes: http://tracker.ceph.com/issues/18476
Signed-off-by: Radoslaw Zarzynski <rzarzynski@mirantis.com>
The call is cognate to removed code in radosgw, and triggers an
assert that the crypto module has already been shutdown. With this
and preceding changes, I see clean shutdown on abnormal termination
of nfs-ganesha during startup.
Fixes: http://tracker.ceph.com/issues/18585
Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
In some librgw shutdown cases, it appears that the shared
object runtime has partially destructed the static map when
our shutdown code is running.
Fixes: http://tracker.ceph.com/issues/18585
Signed-off-by: Matt Benjamin <mbenjamin@redhat.com>
Instead of having a separate output argument with the number of
blocks allocated, just return it via the return value. Simplifies
the calling convention.
Signed-off-by: Sage Weil <sage@redhat.com>
ExtentList was previous relying the caller to preallocate/size the
vector to be large enough for the worst case allocation of extents,
and keeping it's own manual count of the extent list size. Instead,
manage that from ExtentList, and remove the preallocation from the
callers.
Fixes: http://tracker.ceph.com/issues/18573
Signed-off-by: Sage Weil <sage@redhat.com>
os/bluestore: avoid unnecessary memory copy, use variable reference in BlockDevice::Open
Reviewed-by: Sage Weil <sage@redhat.com>
Reviewed-by: Igor Fedotov <ifedotov@mirantis.com>
