#960358 exim4-base: spec.txt, NFS and file locking

Package:
exim4-base
Source:
exim4
Description:
support files for all Exim MTA (v4) packages
Submitter:
Vincent Lefevre
Date:
2026-10-10 11:41:02 UTC
Severity:
normal
Tags:
#960358#5
Date:
2020-05-12 00:17:01 UTC
From:
To:
In /usr/share/doc/exim4-base/spec.txt.gz :
--------
In order to append to an NFS file safely from more than one host, it is
necessary to take out a lock before opening the file, and the lock file
achieves this. Otherwise, even with fcntl() locking, there is a risk of file
corruption.
--------

I don't see why. If used correctly, fcntl() locking should be
sufficient under NFS. And if fcntl() locking is broken, lock files
won't prevent corruption anyway: even if you have the lock, there
is no guarantee that the client will synchronize the changes with
the server before another process gets the lock and do changes
(this is the problem I had with shared zsh history, which often
got corrupted even though zsh used some form of locking... until
I patched it in 2008 to use fcntl() locking).
#960358#10
Date:
2020-05-12 16:57:26 UTC
From:
To:
[...]

Hello Vincent,

Would you mind taking question upstream (exim-users mailing list)?

This does not seem to be a bug report but question that would best be
discussed directly.

cu Andreas
#960358#15
Date:
2020-05-13 21:51:24 UTC
From:
To:
Hi Andreas,

Done 9 hours ago, but it seems that the mailing-list has some issues.

Well, the spec seems to be incorrect. So that's more than a question.
But let's first see what will be said in the exim-users mailing-list.
#960358#20
Date:
2020-05-14 05:40:17 UTC
From:
To:
IMO, Debian is not in a position to change the upstream docs. That's a
thing best done upstream.

Greetings
Marc
#960358#25
Date:
2020-05-14 11:51:11 UTC
From:
To:
Yes, but I don't know what to do. Even though my mail to exim-users
was accepted [status=sent (250 OK id=1jYqvc-0001jm-6G)], I can't see
it on <https://lists.exim.org/lurker/list/exim-users.en.html>.
#960358#30
Date:
2020-05-14 16:16:43 UTC
From:
To:
On 2020-05-14 Vincent Lefevre <vincent@vinc17.net> wrote:
[...]

No worries, iirc exim-users is moderated for non-members. It should go
through.

cu Andreas
#960358#35
Date:
2020-05-14 21:51:22 UTC
From:
To:
Indeed, it had to be approved (note that I subscribed before I posted,
but perhaps this is because this was my first post to the list).

The thread should be available there:

https://lists.exim.org/lurker/message/20200513.125602.7be7983c.en.html
#960358#46
Date:
2026-10-10 11:38:21 UTC
From:
To:
This was fixed in 4.100 with commit
1da9c193f2851ca19edf72ac626d133521586491

Docs: correct NFS file locking advice for appendfile (closes #3147)

The use_lockfile description claimed fcntl() locking risks file corruption
on NFS while a lock file prevents it. As pointed out in the exim-users
thread [1] and Debian #960358 [2], that is unfounded: if fcntl() over NFS
works, it is sufficient; if it does not, a lock file does not rescue the
situation either (client-side caching can still let stale writes reach
the server). Rewrite the paragraph and soften the matching "always use
lock files" advice in the file option description.

cu Andreas