Mount Point goes into a very strange state.

Hi Guys,

This one has got me completely baffled and after some fairly lengthy searching online and in the forum, I think I should share this with you - after all someone is bound to have seen this?

So the story so far, I've built a couple of RHEL 7.5 servers, these are HP DL360 G10's with two six core Xeon, 4 * 300Gb SSD and 64Gb RAM along with 6 network connections in bonds - all good so far.

While building the test server I lost a couple of directories to the same fault and effectively just recreated the box from an image due to time constraints.

However, I now have the problem again and am getting ready to go out in the carpark with the other SA and perform the formation headless chicken dance. The problem shows up as follows on both XFS and in this case a CIFS mount.

Mount the Share;

mount -t cifs -o username=XXXXXX,password=XXXXXX,dir_mode=0775,uid=520,gid=500 //CATHNAS01/SP_DB_EXP /mnt

This bit works and gives;

fbakirpomp2 (root) /-> df -h /mnt
Filesystem             Size  Used Avail Use% Mounted on
//CATHNAS01/SP_DB_EXP  1.5T 1000G  501G  67% /mnt
fbakirpomp2 (root) /->
fbakirpomp2 (root) /-> ls -l / | grep mnt
drwxrwxr-x    1 oraogi  dba         0 Nov 12 16:40 mnt
drwxr-xr-x    2 root    root     4096 Oct  3  2016 mnt1
fbakirpomp2 (root) /->

All good so far;
However after a little while (on the machine with the problem) we get;

[root@fbakirpomp3 ~]# df -h /mnt
df: '/mnt': No such device or address
[root@fbakirpomp3 ~]# ls -l / | grep mnt
ls: cannot access /mnt: No such device or address
d??????????   ? ?       ?          ?            ? mnt
[root@fbakirpomp3 ~]#

So it would seem that the mountpoint has really got itself in a panic, any pointes here would be handy.

The RHEL 7.5 install exhibits this "feature" in the version installed straight from media and the version where I have patched from the RHSM repo with the following command.

yum install yum-plugin-versionlock
  yum --releasever=7.5 update

So if anyone has seen this problem or has any pointers I'd be very grateful.

Regards

Gull04
hist

Yes, well, there have been all sorts of screw ups in the development of these modules as new security features have (tried) to be added.

I think this is probably a known bug.

Have you tried:

mount -t cifs -o username=XXXXXX,password=XXXXXX,dir_mode=0775,uid=520,gid=500,vers=1.0 //CATHNAS01/SP_DB_EXP /mnt

(i.e. using a previous version) to see if it stays up?

If vers=1.0 works then try vers=2.0

Or have you already tried that? You could also (if you are the sysadmin) try this on the share publishing end of things too.

Backing out to previous version(s) is where I would start but perhaps you've already tried that.

Another thought is this. There have also been issues with such mounts resuming activity after hibernation (or link timeout) whereby upon resumption there is a problem authenticating (credentials error). Does this error only occur after a significant inactivity on the share? Does it not happen if you keep it busy? Could be useful to test that out.

Hey.

Normally, on Linux systems, dmesg will show strange things at the H/W level.

Can you post your dmesg output?

Hi Guys,

Some more info on this, but first the dmesg output - well the bit that matters I think.

[49732.732763] CIFS VFS: Send error in QFSAttributeInfo = -6
[49732.735099] CIFS VFS: cifs_mount failed w/return code = -6
[49732.737252] CIFS VFS: Send error in QFSAttributeInfo = -6
[49732.739391] CIFS VFS: cifs_mount failed w/return code = -6
[49744.903644] IPv4: martian source 255.255.255.255 from 10.1.100.205, on dev bond1
[49744.903661] ll header: 00000000: ff ff ff ff ff ff 00 50 56 aa 3e a5 08 00        .......PV.>...
[49777.214179] CIFS VFS: Send error in QFSAttributeInfo = -6
[49777.216321] CIFS VFS: cifs_mount failed w/return code = -6
[49777.218360] CIFS VFS: Send error in QFSAttributeInfo = -6
[49777.220431] CIFS VFS: cifs_mount failed w/return code = -6
[49804.989527] IPv4: martian source 255.255.255.255 from 10.1.100.205, on dev bond1
[49804.989544] ll header: 00000000: ff ff ff ff ff ff 00 50 56 aa 3e a5 08 00        .......PV.>...
[49865.075246] IPv4: martian source 255.255.255.255 from 10.1.100.205, on dev bond1
[49865.075262] ll header: 00000000: ff ff ff ff ff ff 00 50 56 aa 3e a5 08 00        .......PV.>...
[49925.161052] IPv4: martian source 255.255.255.255 from 10.1.100.205, on dev bond1
[49925.161068] ll header: 00000000: ff ff ff ff ff ff 00 50 56 aa 3e a5 08 00        .......PV.>...
[49948.451752] CIFS VFS: Send error in QFSAttributeInfo = -6
[49948.453737] CIFS VFS: cifs_mount failed w/return code = -6
[49948.455770] CIFS VFS: Send error in QFSAttributeInfo = -6
[49948.457747] CIFS VFS: cifs_mount failed w/return code = -6
[49974.570525] CIFS VFS: Send error in QFSAttributeInfo = -6
[49974.572493] CIFS VFS: cifs_mount failed w/return code = -6
[49974.574587] CIFS VFS: Send error in QFSAttributeInfo = -6
[49974.576472] CIFS VFS: cifs_mount failed w/return code = -6

I've currently got this set to version 1.0 for the CIFS type.

I think that I've resolved this, as I said in the first post it happened on my test build server as well and I've made a bit more progress.

Currently I'd like people to consider this thread closed and I'll post a fuller update when I return to the office on Monday. This seems to have turned out to be an LVM issue - which I think was caused by the installer.

I'll post a detailed explanation soon, I'm off to sunny Skye now which is a four hour drive for me if I leave now.

Regards

Gull04