[Samba] Printing: Request to make return connection to client optional samba 3.5.x

[Jeremy Allison]

On Fri, Mar 18, 2011 at 04:19:58PM +0100, Volker Lendecke wrote:
> On Thu, Mar 17, 2011 at 10:57:41AM +0100, Urs Beckmann wrote:
> > We have a network of Samba print-stations. There are complains from
> > users of slow printing, which can be tracked to the
> > return-connection which samba-server makes to print-client for
> > dynamic update of printer-status. Since samba-servers are on
> > different subnets than print-clients, the return-connection from
> > samba to windows-client always fails in windows-firewall, but the
> > tries to build up the return-connection lead to delay in printing
> > and sometimes to delays in printer-properties dialog. It's not
> > possible for us to ask clients to open their firewalls, we need a
> > solution on print-server.
> > What helps is to modify samba source as follows:
> > in file rpc_server/srv_spoolss_nt.c:
> > function spoolss_connect_to_client:
> > do nothing but return False;
> > Unfortunately a modified samba-version is not welcome here
> > politically, so if this feature of optionally (configurable in
> > smb.conf) not making the return-connection could be built into next
> > version of samba in the 3.5.x line it would be very helpful.
> > 
> > An other issue is the timeout for making a connection from Samba to
> > the Windows DC for print authentication, fixed at 10 seconds, which
> > is sometimes too small.
> > This is in file libsmb/cliconnect.c:
> > cli_set_timeout(cli, 10000); /* 10 seconds */
> > If this timeout could optionally be increased to 30 seconds it would
> > also be very helpful.
> 
> Does the attached patch help you? I've added patches for 3.5
> and master. You should add "spoolss backchannel = no" to
> your smb.conf [global] section to disable the backchannel
> for all printers.
> 
> Please report if it solves your issue.
> 
> If you need that in 3.5.next, you should open a bug report
> at https://bugzilla.samba.org.

Missing attachment :-). Can you add it to v3-6-test and
master. I really think we need this one :-).

I'll open the bug report for 3.5.next once you've resent
the attachment.

Thanks !

Jeremy.