rgw: x-amz-date change breaks certain cases of aws sig v4.
Reviewed-by: Matt Benjamin <mbenjamin@redhat.com>
Reviewed-by: Adam C. Emerson <aemerson@redhat.com>
`set_is_rev1()` actually resets the internal state of `FrameAssembler`,
so -- to avoid racing -- it should be performed by the same thread that
takes care about executing `ProtocolV2::write_message()`.
```cppp
void set_is_rev1(bool is_rev1) {
m_descs.clear();
m_flags = 0;
m_is_rev1 = is_rev1;
}
```
See comments in the tracker for analysis.
Fixes: https://tracker.ceph.com/issues/55851
Signed-off-by: Radoslaw Zarzynski <rzarzyns@redhat.com>
crimson/osd: watch list empty return incorrect
Reviewed-by: Chunmei Liu <chunmei.liu@intel.com>
Reviewed-by: Sam Just <sjust@redhat.com>
Reviewed-by: Radoslaw Zarzynski <rzarzyns@redhat.com>
hk.ceph.com - notified mirror maintainer list of mirror decommission on 2020-08-31
us-east.ceph.com - mirror has been offline for years.
Signed-off-by: Matthew Taylor <matthew.taylor@digitalpacific.com.au>
The previous commit moved the entirety of write_image_cache_state()
from under m_lock. This was a step too far because the generated image
cache state json is no longer guaranteed to be consistent.
Arrange for m_lock to still be held during image cache json generation
but released before owner_lock is grabbed.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
This tool iterates through the metadata pool looking for dentries with
non-sensical "first" (snapshot) metadata. It only specifically looks for
damage of the kind where the "first" value is greater than the latest
generated snapshot for the file system. It does not detect other kinds
of genuine damage to the "first" metadata.
Fixes: https://tracker.ceph.com/issues/56140
Signed-off-by: Patrick Donnelly <pdonnell@redhat.com>
The user was not being passed to the created bucket. This means that,
in some operations, radosgw-admin was crashing trying to de-reference
the user.
Fixes: https://tracker.ceph.com/issues/56810
Signed-off-by: Daniel Gryniewicz <dang@redhat.com>
also, let the compiler to deduce the type of the template parameter,
less repeating and less error-prone this way.
Signed-off-by: Kefu Chai <tchaikov@gmail.com>
This came from Josh Salomon:
```
/home/jsalomon/src/prim_score/src/common/Journald.h:93:24: error: ‘SubsystemMap’ does not name a type
93 | JournaldLogger(const SubsystemMap *) {}
| ^~~~~~~~~~~~
/home/jsalomon/src/prim_score/src/common/Journald.h:94:23: error: ‘Entry’ does not name a type
94 | int log_entry(const Entry &) {
|
```
Signed-off-by: Radosław Zarzyński <rzarzyns@redhat.com>
make_writeable keeps track of few stats counters (e.g num_object_clones).
Therfore, we should apply the stats after it.
Signed-off-by: Matan Breizman <mbreizma@redhat.com>
Creating copy-on-write (COW) clones in the backing store when a write arrives
after a snapshot is taken of an object.
Both for Pool or Self-managed snapshotting mechanisms.
Trello: https://trello.com/c/yTDCGNCf
Signed-off-by: Matan Breizman <mbreizma@redhat.com>
Dropping the output of this command entirely, as it is
difficult to maintain it in the docs.A minor change in
the command output invalidates the docs.
Fixes: https://tracker.ceph.com/issues/56162
Signed-off-by: Neeraj Pratap Singh <neesingh@redhat.com>
crimson/{net,osd}: add debugs after bughunt's investigations
Reviewed-by: Samuel Just <sjust@redhat.com>
Reviewed-by: Yingxin Cheng <yingxin.cheng@intel.com>
Reviewed-by: Chunmei Liu <chunmei.liu@intel.com>
