minidlna server not visible in network

[Guido Falsi]

On 12/20/15 09:01, Stefan Esser wrote:
> Am 19.12.2015 um 13:45 schrieb Guido Falsi:
>> On 12/19/15 12:26, Volodymyr Kostyrko wrote:
>>> On 19.12.15 00:03, Dr. Peter Voigt wrote:
>>>> Today I have upgraded net/minidlna to latest version. I am on FreeBSD
>>>> 10.2-RELEASE-p8 (amd64).
>>>>
>>>> Access via web interface on port 8200 is working fine. But the MiniDLNA
>>>> server is not seen in my network and correspondingly I cannot access
>>>> any of my FLAC audio files. I do not host any other media besides the
>>>> FLAC files.
>>>>
>>>> I strongly assume this is related to upgrading to version 1.1.5.
>>>>
>>>> Can anyone reproduce this behavior?
>>>
>>> That's not correct.
>>>
>>> 1. If you already queued any files on your UPnP client whey will play
>>> correctly.
>>> 2. If you leave you UPnP client open for some time it will eventually
>>> find your server.
>>> 3. If your UPnP client is already open restarting minidlna will make it
>>> visible.
>>>
>>> So I'd say it just fails responding to probes.
>>>
>>
>> I posted a reply to the bug report.
>>
>> I'll investigate this one further in thee next few days. I'll followup
>> in the bug report with anything I find out.
> 
> I tried to add some information to the bug report, but bugs.freebsd.org
> denied access and there does not seem to be a way to reset the password.
> (I thought it was the Kerberos password, but that did not work.)
> 

AFAIK it IS the kerberos password. Don't knoww why it's not working for you.

> 
> A sensible test would be a tcpdump of traffic between minidlna server
> and a client, that is started before the minidlna server. My guess is,
> that minidlna replies to a client polling for reachable UPnP servers,
> and that minidlna does not announce its availability when started, but
> only in response to clients polling for a server.

I did a tcpdump, but I also discovered why I could not notice this
problem. I run minidlna in a jail, and jails don't get multicast
packets, so in my case mindlna never replies to those probes anyway,
that's why I lowered the announce timer, which is a good workaround but
not a real solution to the problem.

To the OP, Are you using minidlna in a jail on on the base machine?

I'll repeat the test with minidlna on my base machine soon, so I can
check what's different.

Next step, locating the code which should reply the probes in minidlna
and check for diffs against the previous version.

-- 
Guido Falsi <madpilot at FreeBSD.org>