If we report the DB space as vailable, ceph thinks the OSD can store more
data and will not mark the cluster as full as easily. And in reality, we
can't actually store data in this space--only metadata. Avoid the problem
by not reporting it as available.
Fixes: http://tracker.ceph.com/issues/18599
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit
c66d5babb1e283869ba0f1f59029bead5ca5f37d)
// include dedicated db, too, if that isn't the shared device.
if (bluefs_shared_bdev != BlueFS::BDEV_DB) {
- buf->available += bluefs->get_free(BlueFS::BDEV_DB);
buf->total += bluefs->get_total(BlueFS::BDEV_DB);
}
}