[Spce-user] SPCE on AWS with both internal (VPC) and external endpoints

[Skyler]

Yep. I know the problem well. Solution requires a custom.tt2

If yer a SPCEpro client, then maybe that support can be more assistance. Or
i could be persuaded to share/work on code if you were willing to do the
same, in turn. ;)

--Skyler

On Sep 16, 2016 5:49 AM, "Barry Flanagan" <barry at flanagan.ie> wrote:

On 16 September 2016 at 12:40, Skyler <skchopperguy at gmail.com> wrote:

> Hi,
>
> Ok you got me. I read the first paragraph. Usually when we say 'dev' it
> means "working into production" lol
>

No, we run SPPro in production, but use a CE for testing/dev.

> VPC is a beast on its own. Sorry i gave up on it a while back. You'd
> thinknthey would just 'work' but i couldn't get it done. Thay was to reach
> a paying client even, many hours wasted for nothing.
>

It would all be fine apart from the inability of any hosts in the VPC from
accessing the Elastic IPs assigned to it.

I got RTP working by changing the network.yml settings for eth1 to be
rtp_aws and sip_aws and changing the prefs for this subscriber to set
rtp_interface= 'aws' but still SPCE is sending the public IP in its Route
and Contact header, so call is timing out because SPCE does not see the
ACKs.

-Barry




> --Skyler
>
> On Sep 16, 2016 5:14 AM, "Barry Flanagan" <barry at flanagan.ie> wrote:
>
>> On 16 September 2016 at 11:41, Skyler <skchopperguy at gmail.com> wrote:
>>
>>> You lost me at AWS.
>>>
>>
>> Haha, yeah. I did say it is only a dev box.
>>
>> -Barry
>>
>>
>>> --Skyler
>>>
>>> On Sep 16, 2016 3:47 AM, "Barry Flanagan" <barry at flanagan.ie> wrote:
>>>
>>>> Hi,
>>>>
>>>> We are trying to move our SPCE dev platform to AWS. It works fine for
>>>> external endpoints out of the box, using the advertised_address for the
>>>> main interface.
>>>>
>>>> However, we also have some other endpoints in the same VPC and I can't
>>>> get SPCE to not used the advertised_ip when it is communicating with the
>>>> VPC-internal endpoints - it always advertises its elastic IP.
>>>>
>>>> I have created a second, eth1 interface for use with the VPC hosts, and
>>>> my network.yml is as follows:
>>>>
>>>>     eth0:
>>>>       advertised_ip:
>>>>         - <My Elastic IP>
>>>>       ip: 172.31.24.242
>>>>       netmask: 255.255.240.0
>>>>       type:
>>>>         - web_ext
>>>>         - sip_ext
>>>>         - rtp_ext
>>>>         - web_int
>>>>         - ssh_ext
>>>>     eth1:
>>>>       ip: 172.31.58.136
>>>>       netmask: 255.255.240.0
>>>>       type:
>>>>         - web_ext
>>>>         - sip_ext
>>>>         - rtp_ext
>>>>         - web_int
>>>>         - ssh_ext
>>>>     interfaces:
>>>>       - lo
>>>>       - eth0
>>>>       - eth1
>>>>
>>>>
>>>> SPCE is communicating with the VPC hosts over eth1 as expected, but it
>>>> is advertising the Elastic IP instead of 172.31.58.136. I tried adding an
>>>> advertised_ip of 172.31.58.136 to eth1 as well to force it, but no go.
>>>>
>>>> If I were able to specify a hostname in advertised_ip I could make it
>>>> work by manipulating DNS, but this does not seem to work as rtpengine
>>>> complains "Invalid interface specification: ext"
>>>>
>>>> Any ideas?
>>>>
>>>> Thanks
>>>>
>>>>
>>>> -Barry Flanagan
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> Spce-user mailing list
>>>> Spce-user at lists.sipwise.com
>>>> https://lists.sipwise.com/listinfo/spce-user
>>>>
>>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.sipwise.com/pipermail/spce-user_lists.sipwise.com/attachments/20160916/5a639fc6/attachment-0001.html>