This is the companion kernel patchset to this ceph userland pull req:
https://github.com/ceph/ceph/pull/11710
The problem is that fsync can be very slow on ceph, as it waits for a
cap flush ack. Cap flushes however are generally considered by the MDS
to be background activity, so they don't entail a journal flush on their
own.
The idea here is to add a new flag to cap requests to inform the MDS
that the client is waiting on the reply and that it shouldn't delay it.
I have a small program that opens a file, writes 16k to it, does an
fsync on it and then a close. It does this 256 times, and with the
current code, it takes around 20 minutes to run (~5s for each file).
With this and the companion patch series, it runs in around 12s.
Jeff Layton (5):
ceph: fix minor typo in unsafe_request_wait
ceph: move xattr initialzation before the encoding past the
ceph_mds_caps
ceph: define new argument structure for send_cap_msg
ceph: update cap message struct version to 10
ceph: add flags parameter to send_cap_msg
fs/ceph/caps.c | 289 ++++++++++++++++++++++++++-----------------
include/linux/ceph/ceph_fs.h | 3 +
2 files changed, 177 insertions(+), 115 deletions(-)
--
2.7.4
--
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
