Created attachment 1242218 [details]
Description of problem:
I have project where we tried to use the latest Gluster 3.7.19 cluster for shared storage in our HPC environment. I have a directory structure under /mnt/gluster/ as follows:
In each location I have files as well.
When I'm trying to copy testdir into testcfg2(umask750) with a normal user "who is the owner of the mountpoint and directories" with cp -a it fails after creating subdir.
This is quite important for us as the HPV cluster scripts are doing their copy tasks with preserving the ownership, links etc.
Version-Release number of selected component (if applicable):
OS: RHEL 7.2
Steps to Reproduce:
1.Install Gluster 3.7.19
2.create a volume, enable nfs
3.munt volume remotely
chmod 750 testcfg
echo "eeee" > eee.txt
chmod 550 subdir/
chmod 550 testdir/
chmod 750 testcfg2
cp -a ../testcfg/testdir/ .
cp: cannot create directory `./testdir/subdir': Remote I/O error
ls -all /mnt/gluster/test/testcfg2/testdir/
dr-xr-x---. 2 tibcodev algo 4096 Jan 18 16:13 .
drwxr-x---. 3 tibcodev algo 4096 Jan 18 16:13 ..
-rw-r--r--. 1 tibcodev algo 5 Jan 18 16:13 eee.txt
dr-xr-x---. 3 tibcodev algo 4096 Jan 18 16:14 .
drwxr-x---. 3 tibcodev algo 4096 Jan 18 16:14 ..
-rw-r--r--. 1 tibcodev algo 5 Jan 18 16:14 eee.txt
dr-xr-x---. 2 tibcodev algo 4096 Jan 18 16:14 subdir
I have tested it with glusterfs mounts, the problem does not appeared. Also tested with locally mounted gluster volumes on the gluster host and linux nfs server, the error appears.
Did a test on a normal nfs share, the problem does not appears.
I have reverted back to Gluster version 3.7.11 what I was using during the PoC phase of the project in May 2016. There is no issue like this with that version.
I was not able to choose 3.7.19 in the list of versions during the bug report.
This bug is getting closed because GlusteFS-3.7 has reached its end-of-life.
Note: This bug is being closed using a script. No verification has been performed to check if it still exists on newer releases of GlusterFS.
If this bug still exists in newer GlusterFS releases, please reopen this bug against the newer release.