> Am 18.03.2018 um 20:34 schrieb Gregg Smith <[email protected]>:
> 
> My read on the original post:
> 
> First we have stated that "For mod_ssl to work in the vote release, mod_md 
> must also be included..."
> 
> That is what I honed in on. Apache will not start if there's a module 
> specific directive without that module being loaded. Since the OP states that 
> *mod_ssl* will not work without without mod_md included, there must be some 
> mod_md directives not contained inside <IfModule> laying around in the OP's 
> config. I believe this is the first of two parts.

Exactly. Everything works as before when one does not load mod_md. 

> Now, Apache serving a 404 on /.well-known/acme-challenge/test.txt when mod_md 
> is loaded I think is because mod_md stores this stuff under MDStoreDir where 
> the acme client puts it elsewhere IIRC. So this behavior I see as by design 
> since mod_md intercepts the requests coming from the acme server obviously to 
> serve what is stored under MDStoreDir.
> 
> My guess anyway.

Correct. And as noted in another mail, the fallback behaviour will be added so 
that md and external clients can co-exist.

I did not foresee this mixed run mode and therefore decided to deny any 
fallback here. Seems like this security reduced the usability too much.

Stefan

>> On 3/18/2018 12:07 PM, Eric Covener wrote:
>>> On Sun, Mar 18, 2018 at 2:25 PM, Steffen <[email protected]> wrote:
>>> 
>>> It is indeed a limitation for an "old" account, and when LE enables TLS
>>> again (not sure it does already in ACMEv2 protocol)
>> When did this become about TLS-SNI challenges and how does that tie
>> into the external ACME client?
>> Can you connect the dots for me or is this unrelated?
>>> In my test mod_md says;
>>> 
>>> mod_md.c(1317): [client 2001:980:a510:1:c5e7:56f7:9d:ab36:65315] Challenge
>>> for www.apachelounge.com (/.well-known/acme-challenge/test.txt)
>>> 
>>> 
>>> For me case closed., sorry for the clutter.
>> Does this confirm something beyond "mod_md works"?
>>> When it is not  appreciated that I share it with dev, say it please.
>> My own 2 cents: It would be helpful and take much less of a toll on
>> this volunteers time/patience/morale if this kind of feedback were
>> refined before being brought forward.
>> For example, here are hypothetical concise requirements / complaints
>> that someone could meaningfully address without having to pull teeth:
>> mod_md could do something specifically different with TLS-SNI
>> challenges for old users
>> mod_md pre-empts HTTP challenges for domains that are not mod_md managed.
>> mod_md can't decline/defer to an Alias for /.well-known if it has no
>> stored challenge
>> But instead we have several paragraphs about votes and releases and
>> mod_ssl depending on mod_md and two different clients and a request to
>> test "it" on Linux.

Reply via email to