mkdir -p fails with "no data available" when umask is 0022, but works when umask is 0002.
Also recursive copy (cp -R or cp -r) fails with "no data available", independly of the umask.
See below for an example to reproduce the error. I already tried to change transport from rdma to tcp. (Changing the transport works, but the error persists.)
I'd be grateful for some insight.
I have small test system that still runs on glusterfs 3.12.3, where everything works fine.
[hpcadmin@pcph00131 bug]$ mount | grep gluster
qloginx:/glurch.rdma on /glurch type fuse.glusterfs (rw,relatime,user_id=0,group_id=0,allow_other,max_read=131072)
[hpcadmin@pcph00131 bug]$ pwd
/glurch/test/bug
[hpcadmin@pcph00131 bug]$ umask 0022
[hpcadmin@pcph00131 bug]$ mkdir aa #works
[hpcadmin@pcph00131 bug]$ mkdir -p aa/bb/cc/dd
mkdir: cannot create directory ‘aa/bb’: No data available
[hpcadmin@pcph00131 bug]$ umask 0002
[hpcadmin@pcph00131 bug]$ mkdir -p aa/bb/cc/dd
mkdir: cannot create directory ‘aa/bb’: No data available
[hpcadmin@pcph00131 bug]$ rm -rf aa
[hpcadmin@pcph00131 bug]$ mkdir -p aa/bb/cc/dd #works now
# looks like all directories in the path of mkdir -p ...
# have to be group-writable...
[hpcadmin@pcph00131 bug]$ cp -R aa foo
cp: cannot create directory ‘foo/bb’: No data available
# I cannot get this to work, independ of the umask...
[hpcadmin@pcph00131 bug]$ glusterd --version
glusterfs 3.12.4
[hpcadmin@pcph00131 bug]$ glusterfs --version
glusterfs 3.12.4
[hpcadmin@pcph00131 bug]$ glusterfsd --version
glusterfs 3.12.4
[hpcadmin@pcph00131 bug]$ sudo gluster v status all
Status of volume: glurch
Gluster process TCP Port RDMA Port Online Pid
------------------------------------------------------------------------------
Brick qloginx:/gl/lv1osb06/glurchbrick 0 49152 Y 17623
Brick qloginx:/gl/lv2osb06/glurchbrick 0 49153 Y 17647
Brick gluster2x:/gl/lv3osb06/glurchbrick 0 49152 Y 2845
Brick gluster2x:/gl/lv4osb06/glurchbrick 0 49153 Y 2868
#note: im mounting on the qlogin