mgr/dashboard: SSO not working after REST API versioning
Reviewed-by: Alfonso Martínez <almartin@redhat.com>
Reviewed-by: Ernesto Puerta <epuertat@redhat.com>
mgr/dashboard: Use pipe instead of calling function within template wherever possible
Reviewed-by: Ernesto Puerta <epuertat@redhat.com>
Reviewed-by: Nizamudeen A <nia@redhat.com>
mgr/dashboard: Allow managing iSCSI Initiator after removing from group
Reviewed-by: Stephan Müller <smueller@suse.com>
Reviewed-by: Volker Theile <vtheile@suse.com>
This commit fixes Building wheel for cffi (setup.py) ... error:
c/_cffi_backend.c:15:10: fatal error: ffi.h: No such file or directory
some python packages do not offer precompiled binary packages for aarch64 on PyPI.
only happens on non-amd64 arches, so add [!amd64].
Signed-off-by: JiangYu <lnsyyj@hotmail.com>
this is modeling the seastar::with_lock() method with the exception
that obc is also captured in the finally block.
Signed-off-by: Xuehan Xu <xxhdx1985126@gmail.com>
rgw: avoid expiration early triggering caused by overflow
Reviewed-by: Matt Benjamin <mbenjamin@redhat.com>
Reviewed-by: Abhishek Lekshmanan <abhishek@suse.com>
* refs/pull/38034/head:
lvm/create.py: fix a typo in the help message
Reviewed-by: Jan Fajerski <jfajerski@suse.com>
Reviewed-by: Dimitri Savineau <dsavinea@redhat.com>
Reviewed-by: Rishabh Dave <ridave@redhat.com>
if $release is "dev", then the release being built is not a stable
version.
drop ceph.js, as the web page is generated at build time.
since the last user of "releases.json" is gone, now we can
drop the hooks preparing "releases.json" in conf.py as well.
Signed-off-by: Kefu Chai <kchai@redhat.com>
enable/disable #eol-warning based on "is_release_eol" context variable.
this allows us to generate a more static web page. and pave the road to
a releases.json-free doc build process.
Signed-off-by: Kefu Chai <kchai@redhat.com>
check if the release being built is EOL when building the document, so
no need to use javascript to read releases.json anymore.
Signed-off-by: Kefu Chai <kchai@redhat.com>
* refs/pull/38365/head:
doc/mds: fix the default value for mds_dir_max_commit_size
Reviewed-by: Rishabh Dave <ridave@redhat.com>
Reviewed-by: Jos Collin <jcollin@redhat.com>
Without this fix GCC complains:
```
src/kv/MemDB.cc:179:10: warning: ‘r’ may be used uninitialized in this function [-Wmaybe-uninitialized]
179 | return r;
```
Signed-off-by: Radoslaw Zarzynski <rzarzyns@redhat.com>
The bug here was that if a backfill target has an object-to-delete
placed after the very last object of primary, it will be missed.
Signed-off-by: Radoslaw Zarzynski <rzarzyns@redhat.com>
It discovered an issue with occasional missing deletion
of the very last object on replica if it doesn't exist
on primary.
Signed-off-by: Radoslaw Zarzynski <rzarzyns@redhat.com>
The unittesting discovered an issue between `BackfillState` and
the low-layer recovery code: `RecoveryBackend::recover_object()`.
`BackfillState::enqueue_push()` was assuming it controlls also
to which backfill target the push is sent while `recover_object()`
calculated the set on its own.
Fixing this is the reason why `enqueue_push()` loses the `const
pg_shard_t& target` parameter.
Signed-off-by: Radoslaw Zarzynski <rzarzyns@redhat.com>
This commit makes the backfill code unware about pg log
implementation details. The benefit is easier unit testing.
Signed-off-by: Radoslaw Zarzynski <rzarzyns@redhat.com>
`get_info()` gets replaced with `get_last_update() and
`get_log_tail()`. This entirely removes the awareness of
`pg_info_t` in the backfill code.
Signed-off-by: Radoslaw Zarzynski <rzarzyns@redhat.com>
`get_peer_info()` gets replaced with `get_peer_last_backfill()`.
The benefit is no need to mock peer's `pg_info_t` in unit test.
Signed-off-by: Radoslaw Zarzynski <rzarzyns@redhat.com>
instead of hardwiring the "|stable-release|" to "octopus" use the latest
release name found in 'doc/releases/releases.yml'
Signed-off-by: Kefu Chai <kchai@redhat.com>
