svn commit: samba r20966 - in branches: SAMBA_3_0/source/smbd SAMBA_3_0_24/source/smbd

[James Peach]

On 22/01/2007, at 8:55 PM, Andrew Bartlett wrote:

> On Mon, 2007-01-22 at 20:51 -0800, James Peach wrote:
>> On 22/01/2007, at 8:45 PM, Andrew Bartlett wrote:
>>
>>> On Tue, 2007-01-23 at 04:12 +0000, jpeach at samba.org wrote:
>>>> Author: jpeach
>>>> Date: 2007-01-23 04:12:51 +0000 (Tue, 23 Jan 2007)
>>>> New Revision: 20966
>>>>
>>>> WebSVN: http://websvn.samba.org/cgi-bin/viewcvs.cgi?
>>>> view=rev&root=samba&rev=20966
>>>>
>>>> Log:
>>>> Only attempt to reload the config file atfer the fork point if we
>>>> are in daemon mode. If we are in inetd mode, there's really no  
>>>> point
>>>> in rechecking it so soon.
>>>
>>> Other than the old build farm setup (do we still use that part?) Who
>>> runs smbd in inetd mode?
>>
>> In certain configurations on Mac OS X, it runs from launchd using
>> launchd's inetd compatibility mode. If the samba inetd code was
>> deprecated, I wouldn't be that disappointed, I believe that there's
>> an API where you can check in with launchd after daemonising.
>
> For smbd, I think the issue is the background print queue monitor

You don't get this in inetd mode.

> .  nmbd
> doesn't work to well from inetd, due to the multiple UDP ports etc.

Yup, nmbd gets to run persistently (but still via launchd).

>
> The main point I'll make it that it just isn't tested very often.

Yes, I've been burned before by living off the beaten track :(

>   That
> said, it is nice not to have smbd in memory, if the machine may never
> see a CIFS packet...

That's the primary rationale.

--
James Peach | jpeach at samba.org