Hi, I don't know if this a bug or I am doing something wrong. But it seem that mounting a volume with acl,selinux results in poor performance.
The odd thing is that sometimes it's really slow and other times it just relatively slow. I tried both on glusterfs version 3.6.2 and 3.6.3. The file system on the bricks are xfs created with "-i size=512" and the gluster volume is a "Distributed-Replicate, Bricks: 2 x 2 = 4". HowTo repro # dd if=/dev/zero of=/tmp/masterfile bs=1M count=1000 # shred /tmp/masterfile # mkdir /mnt/selinuxtst /mnt/noselinuxtst # mount -t glusterfs -o rw,acl,selinux localhost:glu_fs /mnt/selinuxtst # mount -t glusterfs -o rw,acl localhost:glu_fs /mnt/noselinuxtst My different results with selinux # time cp /tmp/masterfile /mnt/selinuxtst/master_selinux01 real 0m12.705s user 0m0.007s sys 0m0.660s # time cp /tmp/masterfile /mnt/selinuxtst/master_selinux02 real 7m22.605s user 0m0.025s sys 0m1.776s My different results without selinux # time cp /tmp/masterfile /mnt/noselinuxtst/master_noselinux01 real 0m2.560s user 0m0.007s sys 0m0.661s # time cp /tmp/masterfile /mnt/noselinuxtst/master_noselinux02 real 0m3.588s user 0m0.004s sys 0m0.590s
_______________________________________________ Gluster-users mailing list [email protected] http://www.gluster.org/mailman/listinfo/gluster-users
