[Please advice if a better group exists for this question]

I have three PCs (1 each of Debian-12, Debian-13 and OpenSuse-15.6).
Each PC NFS-mounts a common directory from an NFS-Server (OpenSuse-15.6).

Mount options look the same on all three:

Opensuse:
xxx on xxx type nfs4 (rw,relatime,vers=4.2,rsize=1048576,wsize=1048576,namlen=255,hard,proto=tcp,timeo=600,retrans=2,sec=sys,clientaddr=192.168.1.164,local_lock=none,addr=192.168.1.190)

Debian12
xxx on xxx type nfs4 (rw,relatime,vers=4.2,rsize=1048576,wsize=1048576,namlen=255,hard,proto=tcp,timeo=600,retrans=2,sec=sys,clientaddr=192.168.1.168,local_lock=none,addr=192.168.1.190)

Debian13
xxx on xxx type nfs4 (rw,relatime,vers=4.2,rsize=1048576,wsize=1048576,namlen=255,hard,proto=tcp,timeo=600,retrans=2,sec=sys,clientaddr=192.168.1.179,local_lock=none,addr=192.168.1.190)


When I try to 'flock' a file in a directory on that share simultaneously
on all 3 machines:
% flock foobar sh -c 'echo start ; sleep 10 ; echo end'

- when any Debian-machine holds the lock:
- the other two block as expected

- when the Opensuse machine holds the lock:
- the Debian-12 machine blocks as expected
- the Debian-13 machine fails with
% flock foobar sh -c 'echo start ; sleep 10 ; echo end'
flock: foobar: Input/output error

Any hint why the Debian-13 fails when Opensuse holds the lock?
Any hint where to start to look for answers?

TNX
R'

--- PyGate Linux v1.5.11
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)

Ralf Fassel <ralfixx@gmx.de> writes:

When I try to 'flock' a file in a directory on
that share simultaneously on all 3 machines:

- when the Opensuse machine holds the lock:
- the Debian-12 machine blocks as expected
- the Debian-13 machine fails with
% flock foobar sh -c 'echo start ; sleep 10 ; echo end'
flock: foobar: Input/output error

Any hint why the Debian-13 fails when Opensuse
holds the lock? Any hint where to start to look
for answers?

I would start by trying to build and run
different versions of flock, from Linux utils[1],
by building those versions and seeing if the same
issue holds. Build the newest version, and run it
on all machines, and try to also build the oldest
version you're using and run it on all machines.
If issues seem to get fixed by some combination,
then a bug report against flock might be helpful.
If the same issue still holds, then it probably is
a kernel issue.

If you are up for it, compile and install
different versions of the Linux kernel on those
machines and OSes. See if that would fixes the
issue. In that case you can probably report the
issue against the Linux kernel.

In all cases, you should report the issues
to your distribution's maintainers first, and not
to the Linux upstream directly.

If you are system-programming-inclined, you
can use strace on flock to see what system calls
flock was calling when the error occurred, which
might help you figure out the source of the issue,
with the help of flock's source code.

[1]: https://www.kernel.org/pub/linux/utils/util-linux/


Best Regards,
Lumin Etherlight

--- PyGate Linux v1.5.11
* Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)