I'll correct that right now, for the next release.
> On 13 Oct 2016, at 21:22, Timothy McKernan <[email protected]> wrote:
>
> I was working off of the comments in the local.ini, which suggest using
> {couch_httpd, start_link, [https]}
>
> But your explanation makes it obvious. Thanks.
>
> On Thu, Oct 13, 2016 at 4:00 PM Robert Samuel Newson <[email protected]>
> wrote:
>
>> A common mistake is to use couch_httpd instead of chttpd when configuring
>> the httpsd daemon.
>>
>> This;
>> {couch_httpd, start_link, [https]}
>>
>> will configure a new SSL-enabled listener equivalent to :5986, the
>> node-local port. This port should not bound to anything but 127.0.0.1, so
>> SSL-enabling it makes little sense.
>>
>> This;
>> {chttpd, start_link, [https]}
>>
>>
>> will configure a new SSL-enabled listener equivalent to :5984. This is
>> what you want.
>>
>> B.
>>
>>> On 13 Oct 2016, at 20:34, Timothy McKernan <[email protected]>
>> wrote:
>>>
>>> In my last post you told me (and the docs have always been clear in this
>>> regard) to use port 5984 on my single node, 2.0.0 build.
>>>
>>> If I use Fauxton to complete the single-node setup it asks for "Port that
>>> the Node uses" and defaults to 5984. So far this is all fine.
>>>
>>> I'm using ssl nearly exclusively. It's a requirement for me. The only
>> place
>>> I haven't used it is in using the _setup_cluster endpoint and to test
>>> single node setup using Fauxton.
>>>
>>> If I next PUT a db over SSL/6984 it will only be accessible via ports
>>> 6984/5986, so the SSL daemon appears to be serving the node, not the
>>> cluster.
>>>
>>> Is this intentional? Or did I do something wrong?
>>
>>
