samba - Sep 2023 - sockets in /var/lib

On Tue, 2023-09-12 at 11:07 +0300, Michael Tokarev
wrote:
> Generally, samba does place its files a bit arbitrary and
> unthinkfully.For example, in debian I have a patch which move socket
> directory (whichis definitely a runtime thing, with sockets to
> communicate with runningprocesses) to lock directory instead of
> having them in /var/lib/samba/ (!!)where they don't fit at all.
The main reason that at least some of the sockets are under
/var/lib/samba is that the socket permisisons are long-term state, even
if the listeners are ephemeral.
I agree it feels a bit random, partly due to trying not to break old
installations, but we did read the FHS docs and put in some effort to
sort things into the right categories. 
Andrew Bartlett

-- 
Andrew Bartlett (he/him)       https://samba.org/~abartlet/Samba Team Member
(since 2001) https://samba.orgSamba Team Lead               
https://catalyst.net.nz/services/sambaCatalyst.Net Ltd
Proudly developing Samba for Catalyst.Net Ltd - a Catalyst IT group
company
Samba Development and Support: https://catalyst.net.nz/services/samba
Catalyst IT - Expert Open Source Solutions

12.09.2023 11:40, Andrew Bartlett via samba:
> On Tue, 2023-09-12 at 11:07 +0300, Michael Tokarev wrote:
>> Generally, samba does place its files a bit arbitrary and
>> unthinkfully.For example, in debian I have a patch which move socket
>> directory (whichis definitely a runtime thing, with sockets to
>> communicate with runningprocesses) to lock directory instead of
>> having them in /var/lib/samba/ (!!)where they don't fit at all.
> 
> The main reason that at least some of the sockets are under
> /var/lib/samba is that the socket permisisons are long-term state, even
> if the listeners are ephemeral.
Um.

# ll -a /run/samba/msg.sock/
total 0
drwx------ 2 root root 480 Sep 12 11:43 ./
drwxr-xr-x 7 root root 480 Sep 12 02:50 ../
srwxrwxrwx 1 root root   0 Sep 10 12:28 106srwxrwxrwx 1 root root   0 Sep 10
12:28 107srwxrwxrwx 1 root root   0 Sep 10 12:28 136srwxrwxrwx 1 root root   0
Sep 10 12:28 208srwxrwxrwx 1 root root   0 Sep 10 12:28 212srwxrwxrwx 1 root
root   0 Sep 10 12:28 213srwxrwxrwx 1 root root   0 Sep 10 13:55 2173srwxrwxrwx
1 root root   0 Sep 11 06:14 25036srwxrwxrwx 1 root root   0 Sep 11 06:38
25632srwxrwxrwx 1 root root   0 Sep 11 06:43 25766srwxrwxrwx 1 root root   0 Sep
11 06:43 25771srwxrwxrwx 1 root root   0 Sep 11 06:43 25781srwxrwxrwx 1 root
root   0 Sep 11 06:43 25795srwxrwxrwx 1 root root   0 Sep 11 06:43 25799
These are 100% runtime things and definitely have no long-term
state at all.  Each socket is named after a process pid, and
the sockets are used by smbcontrol to communicate with running
samba processes.

I'm not talking about

srwxrwxrwx 1 root root 0 Sep 10 12:28 /var/lib/samba/winbindd_privileged/pipe
for example, or ntp socket, or something like that - these, which
are used by other subsystems, are kept at the same location in
debian.

I'm talking about the 100% transient runtime thing.
msg.sock is not a place for private directory in /var/lib/samba.

/mjt