[Jack-Devel] Fwd: NetJack2 infinite "Waiting for master..."

PrevNext  Index
DateThu, 13 Jun 2013 20:27:06 +0200
From Antun Horvat <[hidden] at gmail dot com>
To[hidden] at lists dot jackaudio dot org
Follow-UpRuslan N. Marchenko Re: [Jack-Devel] Fwd: NetJack2 infinite "Waiting for master..."
---------- Forwarded message ----------
From: Antun Horvat <[hidden]>
Date: Thu, Jun 13, 2013 at 8:26 PM
Subject: Re: [Jack-Devel] NetJack2 infinite "Waiting for master..."
To: "Ruslan N. Marchenko" <[hidden]>


Hello Ruslan and whole Jack community :)

I think that this patch should certainly be included in the next release
because it seems that mainline NetJack2 module (netmanager) does not scale
well with more complex network configurations.
The thing is that in my opinion, network config that I use is not something
extreme in any way since it is only using standard linux bridge nic that
binds two ethernet cards and few virtual lxc/qemu nics.


Now with that said, I need to ask you (the community), are there some
configuration tips for NetJack2 that could reduce latency between master's
and the slave's playback (audioadapter module).  Two machines are connected
by 100Mbit (duplex) links which by ping have about 200micro seconds round
trip time. On both machines I am using simple motherboard build in audio
output devices.

Thank you in advance



On Tue, Jun 11, 2013 at 8:31 PM, Ruslan N. Marchenko <[hidden]>wrote:

> Hi Antun,
> The patch is actually is being reviewed currently since I've made a pull
> request to the trunk. Pull request is not only to improve handshake but
> also to add IPv6 support, so it takes some time.
> I've mistakenly pressed Reply instead of Group-reply so it went to you
> only, otherwise feel free to comment your experience on the list or I can
> repeat the reply to keep the thread consistent.
> Glad it helped.
> Regards,
> Ruslan
>
>
> 2013/6/11 Antun Horvat <[hidden]>
>
>> Hello Ruslan,
>>
>> first I want to thank you for your help.
>> To answer your question, slave is not multihomed, only master is.
>> I have installed jackd2 (only on the master) from your repo and it worked
>> with a "-a" flag (on slave) that pointed to the master.
>>
>> So if anybody else has the same problem with the "waiting for the
>> master..." state, get yourself a "copy" of  (*ppa:rufferson/snd*) ppa
>> and save ~15h of torcher.
>>
>> one more time ... thx for the help :)
>>
>>
>> On Mon, Jun 10, 2013 at 9:10 PM, Ruslan N. Marchenko <[hidden]
>> > wrote:
>>
>>> Hi Antun,
>>> Am So 09 Jun 2013 22:30:11 CEST schrieb Antun Horvat:
>>>
>>>  Hello,
>>>>
>>>> I have a small problem with NetJack2 setup with two Ubuntu boxes.
>>>> It seems that whatever I do i can not establish master-slave
>>>> connection between these two boxes. I tried to follow various
>>>> tutorials but in all cases the end goal could not be reached and it
>>>> always stops at "Waiting for master..."
>>>>
>>>> By default ubuntu ships with 1.9.9 version that through the process of
>>>> hope i updated to 1.9.9.5 and then to 1.9.10 (git) version. In all
>>>> cases the outcome was the same.
>>>>
>>>> Just to be clear i did the following things on the master and slave
>>>> boxes to initiate the connection:
>>>>
>>>> $master>jackd -d alsa &
>>>> $master>jack_load netmanager
>>>> $slave>jackd -d net
>>>>
>>>> # Then i tried  on the slave, (10.125.113.194) is ip of the master
>>>> $slave>jackd -d net  -a 10.125.113.194 # this does not succeed either
>>>>
>>>>
>>>> The important thing is that when I try to add net agent locally it
>>>> works on both machines.
>>>> $master>jackd -d alsa&;
>>>> $master>jack_load netmanager
>>>> $master>jackd -n test -d net -a 127.0.0.1 # it works
>>>>
>>>> My current understanding is that the there is some kind of issue with
>>>> network since it works locally. But the weird thing is that I am
>>>> working with the slave remotly over the SSH which should imply that
>>>> the boxes are reachable.
>>>> Also I flushed all iptables (filter, nat and masquerade on both
>>>> machines), so there shouldn't be an issue of fw.
>>>>
>>>> Also master can see broadcast packets sent from the slave
>>>> $master> tcpdump udp port 19000 # or tcpdump multicast
>>>>
>>>> .. and I am using linux bridge interface that bridges linux slave box
>>>> and mikrotik internet gateway. netstat -gn lists that netmanager has
>>>> registered multicast ip address  on that bridge interface.
>>>>
>>>>
>>> That could be the problem. Is slave multihomed btw?
>>> The thing is that handshake implementation currently is a bit naive and
>>> fails whenever more complex network setup occurs.
>>> As I'm always saying in this cases - you can try out the build from my
>>> ppa (ppa:rufferson/snd) and explicitly specify interface you want to work
>>> on (-f br0).
>>> It also has more debug entries in various network related points so may
>>> shed more light on where the problem is exactly.
>>>
>>>
>>>  It seems that everything I do is not relevant and it always stops at
>>>> that night haunting message:
>>>> "*Waiting for the master...*"
>>>>
>>>>
>>>> .... so pls. could someone point me towards the master...
>>>>
>>>>
>>>
>>> --
>>> Looking forward to reading yours...
>>>     Ruslan N. Marchenko
>>>
>>
>>
>
>
> --
> Looking forward to reading yours.
>      Ruslan N. Marchenko
>
PrevNext  Index

1371148037.10239_0.ltw:2,a <CAHb5DFY5FBRPq1gXyjgD_+VN8banG4Xc_KSLY86TFAN2t6vWtw at mail dot gmail dot com>