On Mon, Apr 4, 2016 at 11:04 PM, Xiaoxi Chen <superdebuger@xxxxxxxxx> wrote: > Hi, > > I am trying the rados namespace feature on CephFS over 10.1.0, it > works well when I was using the admin keyring. But when I trying to > lock user "cephFS.tenantA" to /tenantA, the writing speed is extremely > slow. > > [Direct] > root@slc5b03c-1gtn:/mnt# dd if=/dev/zero of=./test_big_file of=direct > bs=1M count=20 > 5+0 records in > 5+0 records out > 5242880 bytes (5.2 MB) copied, 30.2971 s, 173 kB/s > > [Buffer] > root@slc5b03c-1gtn:/mnt# dd if=/dev/zero of=./test_big_file_2 bs=1M count=20 > 20+0 records in > 20+0 records out > 20971520 bytes (21 MB) copied, 5.68805 s, 3.7 MB/s > > > cephFS.tenantA was created by: > root@slc5b03c-2ucb:/mnt# ceph auth get-or-create client.cephFS.tenantA > mon 'allow r' mds 'allow rw path=/tenantA' osd 'allow rw pool=mds_data > namespace=tenantA' > [client.cephFS.tenantA] > key = AQAgeAJXhe7aORAAqx46SfsOqeZQue5XoBV8cQ== > > the layout of /tenantA is: > root@slc5b03c-2ucb:/mnt# getfattr -n ceph.dir.layout ./tenantA > # file: tenantA > ceph.dir.layout="stripe_unit=4194304 stripe_count=1 > object_size=4194304 pool=mds_data pool_namespace=tenantA" > > > then try to mount via ceph-fuse like: > root@slc5b03c-1gtn:~# ceph-fuse --id cephFS.tenantA --cluster > slc07_ceph_02 --client_mountpoint /tenantA /mnt > ceph-fuse[94654]: starting ceph client > 2016-04-04 07:42:28.096671 7f65000e2e80 -1 init, newargv = > 0x7f650a680f60 newargc=11 > ceph-fuse[94654]: starting fuse > > Note that, the writing speed with admin keyring looks good > root@slc5b03c-2ucb:/mnt/tenantA# dd if=/dev/zero of=./try oflag=direct > bs=1M count=20 > 20+0 records in > 20+0 records out > 20971520 bytes (21 MB) copied, 0.432546 s, 48.5 MB/s > > And the layout works fine > root@slc5b03c-1gtn:/mnt# rados ls --pool mds_data --namespace tenantA > --id cephFS.tenantA --conf /etc/ceph/slc07_ceph_02.conf > 10000000009.00000000 > 100000003f3.00000001 > 100000003f4.00000000 > 100000003f4.00000002 > 100000003f3.00000000 > 100000003f4.00000003 > 100000003f4.00000004 > 100000003f4.00000001 > > > Is that something bugly in ceph-fuse? > > root@slc5b03c-1gtn:~# dpkg -l | grep fuse > ii ceph-fuse 10.1.0-1trusty > amd64 FUSE-based client for the Ceph distributed file > system > ii fuse 2.9.2-4ubuntu4.14.04.1 > amd64 Filesystem in Userspace > ii libfuse2:amd64 2.9.2-4ubuntu4.14.04.1 > amd64 Filesystem in Userspace (library) > I just tried v10.1.0 and current master (10.1.0-490-gcf5d277), but didn't see this behavior. Please check again. If you still see this behavior, please set debug objecter = 10 debug ms = 10, do some direct-IO write and send the log to us. Thanks Yan, Zheng > > -Xiaoxi > -- > 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 -- 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
