cephadm/box: create osds with podman.
Reviewed-by: Ernesto Puerta <epuertat@redhat.com>
Reviewed-by: Juan Miguel Olmo <jolmomar@redhat.com>
Reviewed-by: Nizamudeen A <nia@redhat.com>
Refine the text in the "Maintenance" section of
doc/radosgw/multisite.rst.
https://tracker.ceph.com/issues/58632
Co-authored-by: Cole Mitchell <cole.mitchell@gmail.com>
Signed-off-by: Zac Dover <zac.dover@proton.me>
More specifically, if we get e.g. a 504 while attempting to download a
.list file, fail the build and log the status code.
Signed-off-by: Zack Cerza <zack@redhat.com>
Refine the English in the first fifty lines of multisite.rst. This is
part of a project whose goal is the creation of a document that will
impart a good aerial view of RGW to a first-time reader in the minimum
number of words.
Co-authored-by: Anthony D'Atri <anthony.datri@gmail.com>
Signed-off-by: Zac Dover <zac.dover@proton.me>
Clean up the English under the section called "Zones". Part of a larger
project aimed at giving the reader more understanding earlier in the
documentation of how buckets and zones and zone groups and endpoints
work.
https://tracker.ceph.com/issues/58632
Signed-off-by: Zac Dover <zac.dover@gmail.com>
It's entirely fine for the map_epoch to change while the op is processed
as long as none of the intervening epochs caused an interval change. In
general, the correct way to check for an interval change is with
has_reset_since.
We don't need to do this check here at all because IOInterruptCondition
will already have performed it against the captured epoch when the
continuation resumed.
Introduced: 31418020fd
Fixes: https://tracker.ceph.com/issues/58486
Signed-off-by: Samuel Just <sjust@redhat.com>
Reverting since this is causing the following ceph-dencoder exception:
[src/tcmalloc.cc:332] Attempt to free invalid pointer 0x558f14e3a000
NOTE: I didn't debug this any more further since there is sizable backlog
of PRs that need testing/merging, but reverting the change makes it work.
Fixes: http://tracker.ceph.com/issues/58219
Signed-off-by: Venky Shankar <vshankar@redhat.com>
mds/Server: do not allow -ve reclaim flags to cause client eviction
Reviewed-by: Rishabh Dave <ridave@redhat.com>
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
Reviewed-by: Xiubo Li <xiubli@redhat.com>
Reviewed-by: Milind Changire <mchangir@redhat.com>
Reviewed-by: Kotresh HR <khiremat@redhat.com>
Force to use the old and slow method to invalidate the dcache
if the euid is non-root, or the remount may fail with return
code 1 or 32.
https://tracker.ceph.com/issues/56532
Signed-off-by: Xiubo Li <xiubli@redhat.com>
The remount will fail randomly for unknown reasons. And in certain
circumstance we can reprodce this very easy, which will block our
testing. Make it posible to force to old method to invalidate the
dcache when the "client_try_dentry_invalidate" option is enabled
even kernel version >= 3.18.0
Fixes: https://tracker.ceph.com/issues/56532
Signed-off-by: Xiubo Li <xiubli@redhat.com>
For some unknown reason the remount will fail, we need to retry it
for the test_dentry_handling.
Fixes: https://tracker.ceph.com/issues/56532
Signed-off-by: Xiubo Li <xiubli@redhat.com>
mds_max_retries_on_remount_failure option is used by Client.cc only.
Fixes: https://tracker.ceph.com/issues/56532
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Add default value for osd_op_thread_suicide_timeout configuration value
to the OSD Configuraton Reference.
https://tracker.ceph.com/issues/58590
Signed-off-by: Zac Dover <zac.dover@gmail.com>
