Description of problem: We have encountered cephfs-fuse mounted directory different behavior than base Linux or kernel driver mounted directories have regarding to the "cd" command against directory with "rw-rw-rw-" permissions: $ id quicklab uid=1000(quicklab) gid=1000(quicklab) groups=1000(quicklab),4(adm) # cat /proc/mounts ... /dev/rbd0 /mnt/ceph-test1 ext4 rw,seclabel,relatime,stripe=1024,data=ordered 0 0 fusectl /sys/fs/fuse/connections fusectl rw,relatime 0 0 ceph-fuse /mnt/ceph-test2 fuse.ceph-fuse rw,nosuid,nodev,relatime,user_id=0,group_id=0,allow_other 0 0 10.74.157.1:6789,10.74.157.55:6789,10.74.156.172:6789:/ /mnt/ceph-test3 ceph rw,relatime,name=admin,secret=<hidden>,acl 0 0 [quicklab@mgmt-0 ~]$ df | grep ceph /dev/rbd0 50264616 5955968 41732264 13% /mnt/ceph-test1 ceph-fuse 111099904 4788224 106311680 5% /mnt/ceph-test2 10.74.157.1:6789,10.74.157.55:6789,10.74.156.172:6789:/ 377286656 33939456 343347200 9% /mnt/ceph-test3 Linux base directory behavior: - in regular Linux you cannot access the directory, as you can with cephfs Fuse: but still can list: [quicklab@mons-0 ~]$ ll /tmp/ total 0 drwxrw-rw-. 2 root root 17 Mar 20 10:44 test [quicklab@mons-0 tmp]$ cd test/ -bash: cd: test/: Permission denied [quicklab@mons-0 ~]$ ll /tmp/test/ ls: cannot access /tmp/test/aaa: Permission denied total 0 -????????? ? ? ? ? ? aaa ----------------------- cephfs with kernel driver: [root@mgmt-0 ~]# ll /mnt/ceph-test3 total 3704832 -rw-r--r-- 1 root root 3793747968 Mar 21 07:58 rhel-server-7.3-x86_64-dvd.iso drwxrw-rw- 1 root root 1 Feb 15 07:19 test drwxr-xr-x 1 root root 21 Mar 21 08:06 var [root@mgmt-0 ~]# cat /mnt/ceph-test3/test/foo bar barbar bar [root@mgmt-0 ~]# logout [quicklab@mgmt-0 ~]$ cat /mnt/ceph-test3/test/foo cat: /mnt/ceph-test3/test/foo: Permission denied [quicklab@mgmt-0 ~]$ ll /mnt/ceph-test3/test/ ls: cannot access /mnt/ceph-test3/test/foo: Permission denied total 0 -????????? ? ? ? ? ? foo [quicklab@mgmt-0 ~]$ cd /mnt/ceph-test3/test/ -bash: cd: /mnt/ceph-test3/test/: Permission denied ----------------------- cephfs_fuse behavior: [quicklab@mgmt-0 ~]$ ll /mnt/ceph-test2/ total 3704833 -rw-r--r--. 1 root root 3793747968 Mar 21 07:58 rhel-server-7.3-x86_64-dvd.iso drwxrw-rw-. 1 root root 15 Feb 15 07:19 test drwxr-xr-x. 1 root root 1107205021 Mar 21 08:06 var [quicklab@mgmt-0 ~]$ ll /mnt/ceph-test2/test/ ls: cannot access /mnt/ceph-test2/test/foo: Permission denied total 0 -????????? ? ? ? ? ? foo -- allow you to get in --- [quicklab@mgmt-0 ~]$ cd /mnt/ceph-test2/test/ <---------------------- [quicklab@mgmt-0 test]$ ll ls: cannot access foo: Permission denied total 0 -????????? ? ? ? ? ? foo [quicklab@mgmt-0 test]$ touch file touch: cannot touch ‘file’: Permission denied Is this behavior expected? Version-Release number of selected component (if applicable): ceph-fuse-12.2.1-40.el7cp.x86_64 How reproducible: always Steps to Reproduce: 1. mount cephfs-fuse directory 2. chmod 666 /mounted/directory 3. cd /mounted/directory Actual results: Expected results: Additional info: client ceph.conf: [quicklab@mgmt-0 ~]$ egrep -v "^#|^$" /etc/ceph/ceph.conf [global] fsid = 5600bcae-f539-4782-83b9-dbccba2b798e mon initial members = mons-0,mons-1,mons-2 mon host = 10.74.157.1,10.74.157.55,10.74.156.172 public network = 10.74.156.0/22 cluster network = 192.168.1.0/ MDS ceph.conf: [root@mdss-1 ~]# egrep -v "^#|^$" /etc/ceph/ceph.conf [global] fsid = 5600bcae-f539-4782-83b9-dbccba2b798e mon host = 10.74.157.1,10.74.157.55,10.74.156.172 public network = 10.74.156.0/22 cluster network = 192.168.1.0/28 [mds.mdss-0] host = mdss-0 [mds.mdss-1] host = mdss-1 # ceph fs status cephfs - 2 clients ====== +------+--------+--------+---------------+-------+-------+ | Rank | State | MDS | Activity | dns | inos | +------+--------+--------+---------------+-------+-------+ | 0 | active | mdss-1 | Reqs: 0 /s | 2261 | 2260 | +------+--------+--------+---------------+-------+-------+ +-----------------+----------+-------+-------+ | Pool | type | used | avail | +-----------------+----------+-------+-------+ | cephfs_metadata | metadata | 9243k | 108G | | cephfs_data | data | 4900M | 108G | +-----------------+----------+-------+-------+ +-------------+ | Standby MDS | +-------------+ | mdss-0 | +-------------+
I'll track this on the following upstream ceph bug for now: http://tracker.ceph.com/issues/23509
MOving this bz to verified state.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2018:1563