Hi,
again, another question that I ran into. Still with my test setup, SLES11
x86_64, lustre 1.8.1.1.
I tried to enable root squash, but I guess I must have done sth. wrong.
I want to enable root squash, with just one exception, therefore on the
MGS/MDT I did:
mgs-mds:/lustre # umount bar
mgs-mds:/lustre # tunefs.lustre --erase-params --param
"mdt.root_squash=65534:65534" --param
"mdt.nosquash_nids=10.0.0.82 at tcp" --
mgsnode="10.0.0.81 at tcp" /dev/xvdb2
checking for existing Lustre data: found CONFIGS/mountdata
Reading CONFIGS/mountdata
Read previous values:
Target: bar-MDT0000
Index: 0
Lustre FS: bar
Mount type: ldiskfs
Flags: 0x441
(MDT update )
Persistent mount opts: errors=remount-ro,iopen_nopriv,user_xattr
Parameters: mgsnode=10.0.0.81 at tcp
Permanent disk data:
Target: bar-MDT0000
Index: 0
Lustre FS: bar
Mount type: ldiskfs
Flags: 0x441
(MDT update )
Persistent mount opts: errors=remount-ro,iopen_nopriv,user_xattr
Parameters: mdt.root_squash=65534:65534 mdt.nosquash_nids=10.0.0.82 at tcp
mgsnode=10.0.0.81 at tcp
Writing CONFIGS/mountdata
and mounted the filesystem again:
mgs-mds:/lustre # mount -t lustre /dev/xvdb2 /lustre/bar-mdt
However, when I run lctl get_param I get this:
mgs-mds:/lustre # lctl get_param mdt.bar-MDT0000.root_squash
error: get_param: /proc/{fs,sys}/{lnet,lustre}/mdt/bar-MDT0000/root_squash:
Found no match
mgs-mds:/lustre # lctl get_param mdt.bar-MDT0000.nosquash_nids
error: get_param: /proc/{fs,sys}/{lnet,lustre}/mdt/bar-MDT0000/nosquash_nids:
Found no match
The value to the parameter mdt.nosquash_nids was taken from lctl list_nids on
the client host, so I guess that should be fine.
Then also mounted the two bar OSTs, and on the clients (10.0.0.89 and
10.0.0.82)
Then still on both clients, I can still delete files as root from files owned
by ordinary users, with permissions like 600.
regards,
Sebastian