Description of problem: cluster.nufa :- if files are created from client mount(not an RHS cluster node) then it should distribute files according to hash value Version-Release number of selected component (if applicable): 3.4.0.15rhs-1.el6rhs.x86_64 How reproducible: Steps to Reproduce: 1. create a Distribute volume and set cluster.nufa on [root@flash ~]# gluster v info nufa Volume Name: nufa Type: Distribute Volume ID: 07444534-2ebf-4115-899d-6dd191826dd8 Status: Started Number of Bricks: 3 Transport-type: tcp Bricks: Brick1: 10.70.43.114:/rhs/brick1/n1 Brick2: 10.70.43.147:/rhs/brick1/n1 Brick3: 10.70.43.115:/rhs/brick1/n1 Options Reconfigured: cluster.nufa: on 2. mount this volume on one of RHS node and on client(should not be RHS node) RHS node:- [root@baratheon ~]# mount | grep nufa 10.70.43.147:/nufa on /mnt/nufa1 type fuse.glusterfs (rw,default_permissions,allow_other,max_read=131072) 10.70.43.147:/nufa on /mnt/nufa1nfs type nfs (rw,addr=10.70.43.147) client:- mount -t glusterfs 10.70.43.147:/nufa /mnt/nufa1 3. create few files from both mount points 4. remove all files from any mount point and set cluster.nufa off and create some files from mount point. set cluster.nufa on. run rebalance once in between 5. repeat step 4 for few times and now make sure cluster.nufa is on and volume does not have any file 6. create few files from client mount point. It should be distributed acroos bricks touch gf{1..10}; mkdir fuse; touch fuse/gf{1..10} [root@baratheon nufa1nfs]# ls -l /rhs/brick1/n1/* -rw-r--r-- 2 root root 0 Aug 6 12:49 /rhs/brick1/n1/gf1 -rw-r--r-- 2 root root 0 Aug 6 12:49 /rhs/brick1/n1/gf10 -rw-r--r-- 2 root root 0 Aug 6 12:49 /rhs/brick1/n1/gf2 -rw-r--r-- 2 root root 0 Aug 6 12:49 /rhs/brick1/n1/gf3 -rw-r--r-- 2 root root 0 Aug 6 12:49 /rhs/brick1/n1/gf4 -rw-r--r-- 2 root root 0 Aug 6 12:49 /rhs/brick1/n1/gf5 -rw-r--r-- 2 root root 0 Aug 6 12:49 /rhs/brick1/n1/gf6 -rw-r--r-- 2 root root 0 Aug 6 12:49 /rhs/brick1/n1/gf7 -rw-r--r-- 2 root root 0 Aug 6 12:49 /rhs/brick1/n1/gf8 -rw-r--r-- 2 root root 0 Aug 6 12:49 /rhs/brick1/n1/gf9 /rhs/brick1/n1/fuse: total 0 -rw-r--r-- 2 root root 0 Aug 6 12:49 gf1 -rw-r--r-- 2 root root 0 Aug 6 12:49 gf10 -rw-r--r-- 2 root root 0 Aug 6 12:49 gf2 -rw-r--r-- 2 root root 0 Aug 6 12:49 gf3 -rw-r--r-- 2 root root 0 Aug 6 12:49 gf4 -rw-r--r-- 2 root root 0 Aug 6 12:49 gf5 -rw-r--r-- 2 root root 0 Aug 6 12:49 gf6 -rw-r--r-- 2 root root 0 Aug 6 12:49 gf7 -rw-r--r-- 2 root root 0 Aug 6 12:49 gf8 -rw-r--r-- 2 root root 0 Aug 6 12:49 gf9 Actual results: all files are stored on one brick (where we have mounted volume) Expected results: If user creates files from mount point (which is other than RHS node) files should be distibuted according to their hash value Additional info:
Created attachment 783667 [details] log
Created attachment 783668 [details] log
Created attachment 783669 [details] log
Created attachment 783670 [details] log
Created attachment 783671 [details] log
Rachana tried to reproduce in the new build. But unable to reproduce. Marking the bug as closed for now.