[asterisk-dev] DNS & PJSIP

Matthew Jordan mjordan at digium.com
Mon Mar 17 09:00:35 CDT 2014


On Mon, Mar 17, 2014 at 4:47 AM, Olle E. Johansson <oej at edvina.net> wrote:
> Friends,
> After a bit of heated discussion I can try to make a conclusion from what I see when the dust settled:
>
> - We are still in control of our own product and make our own decisions about Asterisk architecture. Any arguments like "PJSIP has it so we have to enable it" falls to the ground as not valid and disappear in a cloud of smoke.
>
> - No one has given any reasons why we should be able to configure DNS servers in the PJSIP channel configuration, apart from Jared who wanted to let users shoot themself in the feet. That is not the way I treat my users. I think most of us now agree that we don't want to have that configuration item.
>
> - In the long run, having a DNS resolver embedded in a module is not a good thing. Due to the PJSIP architecture, it's very hard to avoid. I've spoken with several non-asterisk developers using PJSIP that have partly succeeded, but not fully. We need to stress this to pjsip so that they can separate this code in the future.
>
> - Having PJSIP parse /etc/resolv.conf is not a good thing. If that can't be avoided, we need to monitor the file for updates from Asterisk and reconfigure.
>
> Sorry for being a bit provocative, but some responses I got caused me to become upset. I feel very strongly for keeping our architecture in control and following the soul of Asterisk and the environment we run Asterisk in.
>
> I think that's all I see in the dust on the battleground. Any comments on this?
>

I've already spelled out my disagreements with your arguments here
previously [1] [2] [3] [4]. I don't believe rehashing them again is
going to be productive.

Josh has removed the ability for an end user to configure the
nameservers [5]. Asterisk will use the system ones if available - if
those are not available, then the code will revert back to
'gethostbyname' style resolution.

I think this is one of those situations in which we are going to have
to agree to disagree. Disagreement is okay: being able to disagree
means we can have healthy discussions. In this case, this is a change
that I feel is immensely beneficial for end users, and which has
minimal negative implications both for the architecture of Asterisk as
well as for end users. You disagree with that; I respect your opinion.
We are, however, going to move forward with this change.

That does not mean the story is closed on DNS resolution in Asterisk
or in chan_pjsip. As I said previously, I look forward to _any_ patch
that improves DNS resolution capabilities in Asterisk. My fervent hope
is that the DNS resolution capabilities in the core of Asterisk will
reach a point where we can remove any reliance on the resolver in
PJSIP. Doing so will be transparent to end users: as this change now
has zero configuration capabilities, there will be no change from an
end user's perspective on how resolution occurs. I certainly look
forward to such a contribution from the Asterisk community.

[1] http://lists.digium.com/pipermail/asterisk-dev/2014-March/065968.html
[2] http://lists.digium.com/pipermail/asterisk-dev/2014-March/065990.html
[3] http://lists.digium.com/pipermail/asterisk-dev/2014-March/066033.html
[4] http://lists.digium.com/pipermail/asterisk-dev/2014-March/065992.html
[5] http://lists.digium.com/pipermail/asterisk-dev/2014-March/066115.html

-- 
Matthew Jordan
Digium, Inc. | Engineering Manager
445 Jan Davis Drive NW - Huntsville, AL 35806 - USA
Check us out at: http://digium.com & http://asterisk.org



More information about the asterisk-dev mailing list