ceph-create-keys should not be started on boot of mons with systemd so should
not exist as 'After' or 'Wants' for the ceph-mon.service
Signed-off-by: Owen Synge <osynge@suse.com>
ceph-create-keys should not be started on boot of mons with systemd so should
not exist as 'After' or 'Wants' for the ceph-mon.service
Signed-off-by: Owen Synge <osynge@suse.com>
ceph-create-keys should not be started on boot of mons with systemd so should
not exist in the systemd files
Signed-off-by: Owen Synge <osynge@suse.com>
ceph-create-keys should not be started on boot of mons with upstart so should
not exist in the upstart files
Signed-off-by: Owen Synge <osynge@suse.com>
We are going to take a different path (rollback support in ObjectStore
instead of roll-forward via splicing data from a temp object).
Signed-off-by: Sage Weil <sage@redhat.com>
This field is needed by the client when it
receives an MClientCaps from the server, but
it is never read on the serevr when set by
the client.
This change is aimed at making the code easier
to understand by removing something redundant (it's
not a performance optimisation!)
Signed-off-by: John Spray <john.spray@redhat.com>
The intent of that flag is to enable admins to limit a client
to setting layouts to put objects within the OSD auth caps that
it has been given, so we should apply it to namespace as well
as pool ID.
Fixes: http://tracker.ceph.com/issues/17798
Signed-off-by: John Spray <john.spray@redhat.com>
msg/async: ibverbs/rdma support
Reviewed-by: Sage Weil <sage@redhat.com>
Reviewed-by: Mellanox <adirl@mellanox.com>
Reviewed-by: Matt Benjamin <mbenjami@redhat.com>
Reviewed-by: Mark Nelson <mnelson@redhat.com>
Previously running the script as unprivileged user was not very useful
due to difficulty to change path sudo was looking for a command to
execute.
Signed-off-by: Mykola Golub <mgolub@mirantis.com>
The "additional details" section was getting longer
and longer, and I don't want developer-only stuff
in there anyway.
Signed-off-by: John Spray <john.spray@redhat.com>
Some files had multiple top-level headers resulting in
subsections being listed in the TOC alongside the
main page header.
Signed-off-by: John Spray <john.spray@redhat.com>
...and add a warning at the very top of the document
that it is not for ordinary user use.
Considered just not linking this in at all, but it
seems a shame to have it here but not be able to
e.g. google it and link to it.
Signed-off-by: John Spray <john.spray@redhat.com>
So that a user has a natural way of undoing a setxattr
which set a pool_namespace.
Fixes: http://tracker.ceph.com/issues/17797
Signed-off-by: John Spray <john.spray@redhat.com>
Since 6244755a70 and
12e508313d, the original check added in
ae0d6eb1c0 (perhaps by a bad
rebase/merge?) is duplicated and doubles the warning messages.
Signed-off-by: Josh Durgin <jdurgin@redhat.com>
This forces users upgrading past jewel to first set
'sortbitwise', which in turn means that post-jewel OSDs
do not need to support nibblewise sort.
Signed-off-by: Sage Weil <sage@redhat.com>
