This will allow a single-step upgrade from msgr1 to msgr2: as each
nautilus OSD starts up, it will bind to both msgr1 and msgr2 addrs
(assuming it starts up after the mons have been upgraded). Peer OSDs will
use the msgr2 address if they can, or the legacy one if they can't.
Once require_osd_release >= nautilus OSDs will stop binding to msgr1
ports at all.
Signed-off-by: Sage Weil <sage@redhat.com>
Minimal changes to SimpleMessenger and AsyncMessenger to keep things
working, assuming we ony have a single addr in the addrvec.
Signed-off-by: Sage Weil <sage@redhat.com>
This allows users to simply pass in CLUSTER or PUBLIC and the function
fills in the rest. (Or, if the caller is particular, they can pass in
the other flags too.)
Signed-off-by: Sage Weil <sage@redhat.com>
Flags for everything we would want. The intent is to return an addrvec
of addrs that we can directly bind to.
Signed-off-by: Sage Weil <sage@redhat.com>
* refs/pull/22165/head:
qa: add one-off clusters to qa/cephfs/clusters
qa: allocate more space for VM disk
qa/cephfs/clusters/*: bigger cinder volumses
Reviewed-by: Patrick Donnelly <pdonnell@redhat.com>
Now that we have a nice error message when a ceph.conf is missing
and we try to use values from it, maintaining a list of commands that
don't need ceph.conf isn't as helpful. We had actually missed 'simple
trigger' when we first implemented this causing all our luminous tests
for simple to fail when we backported.
Signed-off-by: Andrew Schoen <aschoen@redhat.com>
This setting tends to confuse people, as it's only respected
on the very first startup of the cluster. Instead, mention
it (with appropriate caveats) on the general mgr admin
page.
Signed-off-by: John Spray <john.spray@redhat.com>
If the ceph configuration file is not loaded correctly and then values
from it are used then an undescript error message is shown, e.g.
AttributeError: 'property' object has no attribute 'get'
With this change that same error condition shows the message:
RuntimeError: No valid ceph configuration file was loaded.
Signed-off-by: Andrew Schoen <aschoen@redhat.com>
mgr/dashboard: add supported flag information to config options documentation
Reviewed-by: Sebastian Krah <skrah@suse.com>
Reviewed-by: Tiago Melo <tmelo@suse.com>
to take the full advantage of seastar reactor, we need to confine the
alien threads to given CPU core(s). in current setting, it's assumed
that alien threads do not perform CPU-bound tasks, so a single core
would suffice. we can always change the interface to pass a cpu_set_t
or a std::bitset if one CPU core is not enough.
Signed-off-by: Kefu Chai <kchai@redhat.com>
Performance evaluations of medium to large size Ceph clusters have
demonstrated negligible performance impact from unnecessarily deep
directory hierarchies but significant performance impact from filestore
split and merge activity. Disable merges by default.
Fixes: http://tracker.ceph.com/issues/24686
Signed-off-by: Douglas Fuller <dfuller@redhat.com>
If "/usr/bin/rbdmap unmap-all" notices a file system mounted on top of
an rbd device, it will call umount, interfering with systemd shutdown
logic. Make sure we aren't invoked until all _netdev mounts are dealt
with by systemd.
Fixes: http://tracker.ceph.com/issues/24713
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
