Description of problem: ======================= If Re-balance fails while migrating the file, T bit is getting set on the file, though user may not find any changes from the mount point Version-Release number of selected component (if applicable): ========== glusterfs-api-3.7.1-6 Steps to Reproduce: ========================== 1. Create a distributed volume with two bricks and mount it on client using FUSE 2. Create 4 GB file on mount point 3. Remove the brick which having the file, and while re-balance in progress kill the brick process 4. Bring up the brick process and again do the remove-brick operation and after completion of operation do the ls -lrt on the destination brick, and T bit is set on the files [root@rhs-client9 data3]# ls -lrt total 10223632 -rw-rwsr-T+ 2 root root 3489661429 Sep 10 17:06 file1 -rw-rwsr-T+ 2 root root 6979322366 Sep 10 17:06 file2 Expected Result: =================== T bit should not be set Notes: ========= [root@rhs-client9 data3]# gluster vol status ACL_DHT Status of volume: ACL_DHT Gluster process TCP Port RDMA Port Online Pid ------------------------------------------------------------------------------ Brick rhs-client9.lab.eng.blr.redhat.com:/r hs/brick4/ACL_DHT 49199 0 Y 7345 NFS Server on localhost 2049 0 Y 7975 NFS Server on rhs-client39.lab.eng.blr.redh at.com 2049 0 Y 7187 Task Status of Volume ACL_DHT ------------------------------------------------------------------------------ There are no active volume tasks
One reason for seeing T bit could be that just before removing the linkto xattr from the file the rebalance process failed and hence the T bit was never removed. Rajesh can you recreate the bug and confirm that the files had the T bit after the first rebalance failed?