We ripped out support for preferred osd placement in ceph since it was a
broken misfeature that nobody used. When I went to mirror those changes on
the kernel side I found that a whole bunch of old stuff had never been moved
over, including a number of cleanups and an actual bug fix. Everything is
now in the wip-crush tree on github.com/ceph/ceph-client.
This code is well tested in the sense that it has been this way on the
userland side for ages (with the exception of the first preferred_osds
patch). That said, I've only done very basic testing of the kernel code,
so it's the crush callers that need attention.
I'd like to get this into the testing branch soon so we can see if it
sorts out #2022. A few of these should probably also go to Linus before
the next window, too.
sage
Sage Weil (11):
ceph: drop support for preferred_osd pgs
ceph: refactor SETLAYOUT and SETDIRLAYOUT ioctl checks into common
helper
crush: clean up types, const-ness
crush: adjust local retry threshold
crush: be more tolerant of nonsensical crush maps
crush: use a temporary variable to simplify crush_do_rule
crush: remove forcefeed functionality
crush: remove parent maps
crush: fix tree node weight lookup
crush: fix memory leak when destroying tree buckets
crush: warn on do_rule failure
drivers/block/rbd.c | 1 -
fs/ceph/file.c | 1 -
fs/ceph/ioctl.c | 89 ++++++++++-------------
fs/ceph/xattr.c | 9 ---
include/linux/ceph/ceph_fs.h | 4 +-
include/linux/ceph/osdmap.h | 2 -
include/linux/crush/crush.h | 18 ++---
include/linux/crush/mapper.h | 7 +-
net/ceph/crush/crush.c | 45 +++---------
net/ceph/crush/mapper.c | 162 +++++++++++++++++------------------------
net/ceph/osdmap.c | 67 +++++------------
11 files changed, 147 insertions(+), 258 deletions(-)
--
1.7.9
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
