You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Jim Jagielski <ji...@jaguNET.com> on 2015/09/25 16:31:58 UTC

mod_h2 CTR (Was: Re: svn commit: r1705257 - /httpd/httpd/trunk/modules/http2/config.m4)

I'd like to propose that we consider mod_h2 in 2.4.x as CTR and
in the same category as mod_lua... That is, somewhat "prod-experimental"
in that some features may still change, but stable enough to be
usable in production environs.

Re: AW: mod_h2 CTR (Was: Re: svn commit: r1705257 - /httpd/httpd/trunk/modules/http2/config.m4)

Posted by Gregg Smith <gl...@gknw.net>.
On 9/25/2015 8:56 AM, Plüm, Rüdiger, Vodafone Group wrote:
>
>> -----Ursprüngliche Nachricht-----
>> Von: Plüm, Rüdiger, Vodafone Group
>> Gesendet: Freitag, 25. September 2015 17:51
>> An: dev@httpd.apache.org
>> Betreff: AW: mod_h2 CTR (Was: Re: svn commit: r1705257 -
>> /httpd/httpd/trunk/modules/http2/config.m4)
>>
>>
>>
>>> -----Ursprüngliche Nachricht-----
>>> Von: Jim Jagielski [mailto:jim@jaguNET.com]
>>> Gesendet: Freitag, 25. September 2015 17:46
>>> An: dev@httpd.apache.org
>>> Betreff: Re: mod_h2 CTR (Was: Re: svn commit: r1705257 -
>>> /httpd/httpd/trunk/modules/http2/config.m4)
>>>
>>> In any case, let's not bike-shed on what we call it, or whatever...
>>>
>>> Let's focus on deciding if we are cool with it being CTR once it's
>>> folded into 2.4.x and "handling" it as we do mod_lua.
>> +1 on CTR for mod_h2 in 2.4.x
>>
> Just to clarify: CTR only on the code in modules/http2. All changes / adjustments that might be needed to mod_ssl should stay RTC.
+1 as detailed by Rüdiger.
+1 to experimental and same note as mod_lua in docs.

Re: mod_h2 CTR (Was: Re: svn commit: r1705257 - /httpd/httpd/trunk/modules/http2/config.m4)

Posted by Jim Jagielski <ji...@jaguNET.com>.
I don't think this needs a formal vote, but I'll give it the
weekend, for people to chime in...

Re: mod_h2 CTR (Was: Re: svn commit: r1705257 - /httpd/httpd/trunk/modules/http2/config.m4)

Posted by Eric Covener <co...@gmail.com>.
On Fri, Sep 25, 2015 at 3:53 PM, Greg Stein <gs...@gmail.com> wrote:
> IMO, call it experimental, and detail what that means in the release notes
> or an FAQ on the site.


+1

-- 
Eric Covener
covener@gmail.com

Re: mod_h2 CTR (Was: Re: svn commit: r1705257 - /httpd/httpd/trunk/modules/http2/config.m4)

Posted by Greg Stein <gs...@gmail.com>.
On Fri, Sep 25, 2015 at 11:51 AM, Rainer Jung <ra...@kippdata.de>
wrote:

> Am 25.09.2015 um 17:56 schrieb Plüm, Rüdiger, Vodafone Group:
>
>> ...
>>> +1 on CTR for mod_h2 in 2.4.x
>>>
>>>
>> Just to clarify: CTR only on the code in modules/http2. All changes /
>> adjustments that might be needed to mod_ssl should stay RTC.
>>
>
> +1 as detailed by Rüdiger.
>

+1

IMO, call it experimental, and detail what that means in the release notes
or an FAQ on the site.

Cheers,
-g

Re: mod_h2 CTR (Was: Re: svn commit: r1705257 - /httpd/httpd/trunk/modules/http2/config.m4)

Posted by Rainer Jung <ra...@kippdata.de>.
Am 25.09.2015 um 17:56 schrieb Plüm, Rüdiger, Vodafone Group:
>> -----Ursprüngliche Nachricht-----
>> Von: Plüm, Rüdiger, Vodafone Group
>> Gesendet: Freitag, 25. September 2015 17:51
>> An: dev@httpd.apache.org
>> Betreff: AW: mod_h2 CTR (Was: Re: svn commit: r1705257 -
>> /httpd/httpd/trunk/modules/http2/config.m4)
>>
>>
>>
>>> -----Ursprüngliche Nachricht-----
>>> Von: Jim Jagielski [mailto:jim@jaguNET.com]
>>> Gesendet: Freitag, 25. September 2015 17:46
>>> An: dev@httpd.apache.org
>>> Betreff: Re: mod_h2 CTR (Was: Re: svn commit: r1705257 -
>>> /httpd/httpd/trunk/modules/http2/config.m4)
>>>
>>> In any case, let's not bike-shed on what we call it, or whatever...
>>>
>>> Let's focus on deciding if we are cool with it being CTR once it's
>>> folded into 2.4.x and "handling" it as we do mod_lua.
>>
>> +1 on CTR for mod_h2 in 2.4.x
>>
>
> Just to clarify: CTR only on the code in modules/http2. All changes / adjustments that might be needed to mod_ssl should stay RTC.

+1 as detailed by Rüdiger.

Rainer

Re: mod_h2 CTR (Was: Re: svn commit: r1705257 - /httpd/httpd/trunk/modules/http2/config.m4)

Posted by Jim Jagielski <ji...@jaguNET.com>.
> 
> Just to clarify: CTR only on the code in modules/http2. All changes / adjustments that might be needed to mod_ssl should stay RTC.
> 

Yes.

AW: mod_h2 CTR (Was: Re: svn commit: r1705257 - /httpd/httpd/trunk/modules/http2/config.m4)

Posted by Plüm, Rüdiger, Vodafone Group <ru...@vodafone.com>.

> -----Ursprüngliche Nachricht-----
> Von: Plüm, Rüdiger, Vodafone Group
> Gesendet: Freitag, 25. September 2015 17:51
> An: dev@httpd.apache.org
> Betreff: AW: mod_h2 CTR (Was: Re: svn commit: r1705257 -
> /httpd/httpd/trunk/modules/http2/config.m4)
> 
> 
> 
> > -----Ursprüngliche Nachricht-----
> > Von: Jim Jagielski [mailto:jim@jaguNET.com]
> > Gesendet: Freitag, 25. September 2015 17:46
> > An: dev@httpd.apache.org
> > Betreff: Re: mod_h2 CTR (Was: Re: svn commit: r1705257 -
> > /httpd/httpd/trunk/modules/http2/config.m4)
> >
> > In any case, let's not bike-shed on what we call it, or whatever...
> >
> > Let's focus on deciding if we are cool with it being CTR once it's
> > folded into 2.4.x and "handling" it as we do mod_lua.
> 
> +1 on CTR for mod_h2 in 2.4.x
> 

Just to clarify: CTR only on the code in modules/http2. All changes / adjustments that might be needed to mod_ssl should stay RTC.

Regards

Rüdiger

AW: mod_h2 CTR (Was: Re: svn commit: r1705257 - /httpd/httpd/trunk/modules/http2/config.m4)

Posted by Plüm, Rüdiger, Vodafone Group <ru...@vodafone.com>.

> -----Ursprüngliche Nachricht-----
> Von: Jim Jagielski [mailto:jim@jaguNET.com]
> Gesendet: Freitag, 25. September 2015 17:46
> An: dev@httpd.apache.org
> Betreff: Re: mod_h2 CTR (Was: Re: svn commit: r1705257 -
> /httpd/httpd/trunk/modules/http2/config.m4)
> 
> In any case, let's not bike-shed on what we call it, or whatever...
> 
> Let's focus on deciding if we are cool with it being CTR once it's
> folded into 2.4.x and "handling" it as we do mod_lua.

+1 on CTR for mod_h2 in 2.4.x

Regards

Rüdiger

Re: mod_h2 CTR (Was: Re: svn commit: r1705257 - /httpd/httpd/trunk/modules/http2/config.m4)

Posted by Jim Jagielski <ji...@jaguNET.com>.
In any case, let's not bike-shed on what we call it, or whatever...

Let's focus on deciding if we are cool with it being CTR once it's
folded into 2.4.x and "handling" it as we do mod_lua.

Re: mod_h2 CTR (Was: Re: svn commit: r1705257 - /httpd/httpd/trunk/modules/http2/config.m4)

Posted by Jim Jagielski <ji...@jaguNET.com>.
> On Sep 25, 2015, at 11:09 AM, Eric Covener <co...@gmail.com> wrote:
> 
> On Fri, Sep 25, 2015 at 11:06 AM, Yann Ylavic <yl...@gmail.com> wrote:
>> On Fri, Sep 25, 2015 at 4:31 PM, Jim Jagielski <ji...@jagunet.com> wrote:
>>> I'd like to propose that we consider mod_h2 in 2.4.x as CTR and
>>> in the same category as mod_lua... That is, somewhat "prod-experimental"
>>> in that some features may still change, but stable enough to be
>>> usable in production environs.
>> 
>> I like "prod-experimental" because I think it is production ready, and
>> still it lets us optimize the interface without giving developpers
>> (not users!) strong garanties.
>> 
>> Maybe the ALPN and protocol selection hooks could be marked as such
>> too, we could have the need to change a RUN_ONCE to a RUN_ALL for
>> those and shouldn't give too much garanties to other modules.
>> 
>> So if we can find a good wording to also express "prod-ready" (I guess
>> it already runs successfully in some prods), I'm +1.
> 
> 
> Not to be a wet blanket, but I don't think we have any real data for
> the production part of that. We could say the interfaces,
> configuration and defaults are experimental in the volatile sense.


'prod-dev' ? :)

Re: mod_h2 CTR (Was: Re: svn commit: r1705257 - /httpd/httpd/trunk/modules/http2/config.m4)

Posted by Yann Ylavic <yl...@gmail.com>.
On Fri, Sep 25, 2015 at 5:09 PM, Eric Covener <co...@gmail.com> wrote:
> On Fri, Sep 25, 2015 at 11:06 AM, Yann Ylavic <yl...@gmail.com> wrote:
>> On Fri, Sep 25, 2015 at 4:31 PM, Jim Jagielski <ji...@jagunet.com> wrote:
>>> I'd like to propose that we consider mod_h2 in 2.4.x as CTR and
>>> in the same category as mod_lua... That is, somewhat "prod-experimental"
>>> in that some features may still change, but stable enough to be
>>> usable in production environs.
>>
>> I like "prod-experimental" because I think it is production ready, and
>> still it lets us optimize the interface without giving developpers
>> (not users!) strong garanties.
>>
>> Maybe the ALPN and protocol selection hooks could be marked as such
>> too, we could have the need to change a RUN_ONCE to a RUN_ALL for
>> those and shouldn't give too much garanties to other modules.
>>
>> So if we can find a good wording to also express "prod-ready" (I guess
>> it already runs successfully in some prods), I'm +1.
>
>
> Not to be a wet blanket, but I don't think we have any real data for
> the production part of that. We could say the interfaces,
> configuration and defaults are experimental in the volatile sense.

Not mentioning it is not production ready is fine too :)
I wonder if "experimental" does not sound too much like this...

Re: mod_h2 CTR (Was: Re: svn commit: r1705257 - /httpd/httpd/trunk/modules/http2/config.m4)

Posted by Stefan Eissing <st...@greenbytes.de>.
> Am 25.09.2015 um 17:09 schrieb Eric Covener <co...@gmail.com>:
> 
> On Fri, Sep 25, 2015 at 11:06 AM, Yann Ylavic <yl...@gmail.com> wrote:
>> On Fri, Sep 25, 2015 at 4:31 PM, Jim Jagielski <ji...@jagunet.com> wrote:
>>> I'd like to propose that we consider mod_h2 in 2.4.x as CTR and
>>> in the same category as mod_lua... That is, somewhat "prod-experimental"
>>> in that some features may still change, but stable enough to be
>>> usable in production environs.
>> 
>> I like "prod-experimental" because I think it is production ready, and
>> still it lets us optimize the interface without giving developpers
>> (not users!) strong garanties.
>> 
>> Maybe the ALPN and protocol selection hooks could be marked as such
>> too, we could have the need to change a RUN_ONCE to a RUN_ALL for
>> those and shouldn't give too much garanties to other modules.
>> 
>> So if we can find a good wording to also express "prod-ready" (I guess
>> it already runs successfully in some prods), I'm +1.
> 
> 
> Not to be a wet blanket, but I don't think we have any real data for
> the production part of that. We could say the interfaces,
> configuration and defaults are experimental in the volatile sense.

I am unable to vote no this since the exact definition of the tag
names are not know to me. I have a an idea though, of course...

That being said: The new protocols hooks and the coupling to mod_ssl
could very well need fine tuning. We have feedback from the mod_websocket
maintainer that that module would need some additions. So marking these
parts as "experimental" sounds prudent.

As to the production readiness, waters are much more murky:
- several peeps have it running. thx to them, we found some bugs, which
  we otherwise would not have caught that easily.
- we will continue to find bugs. httpd lives in a huge ecosystem and
  mod_h2 has not seen all sorts of weather yet.
- oth when I look back to what I have seen in enterprises deployed
  as "production ready", it looks very solid.
- As Yann mentioned, httpd stays able to rock-solid serve HTTP
  traffic from clients with mod_h2 unloaded. We have not implemented
  any APIs from mod_h2 to other modules (for example the Server Push)
  feature that would get lost when unloaded.
- Any hot fixes to mod_h2 itself, can be delivered to adventuring
  people via the github repo. With a 2.4.18 httpd out there, peeps
  can build mod_h2 for themselves so much more easily, basically
  just needing the proper apxs. That can give us fast turnaround
  on tests in deployed settings which we merge back to trunk, 
  aggregate and backport for the next 2.4.x

Cheers,

  Stefan

<green/>bytes GmbH
Hafenweg 16, 48155 Münster, Germany
Phone: +49 251 2807760. Amtsgericht Münster: HRB5782

Re: mod_h2 CTR (Was: Re: svn commit: r1705257 - /httpd/httpd/trunk/modules/http2/config.m4)

Posted by Eric Covener <co...@gmail.com>.
On Fri, Sep 25, 2015 at 11:06 AM, Yann Ylavic <yl...@gmail.com> wrote:
> On Fri, Sep 25, 2015 at 4:31 PM, Jim Jagielski <ji...@jagunet.com> wrote:
>> I'd like to propose that we consider mod_h2 in 2.4.x as CTR and
>> in the same category as mod_lua... That is, somewhat "prod-experimental"
>> in that some features may still change, but stable enough to be
>> usable in production environs.
>
> I like "prod-experimental" because I think it is production ready, and
> still it lets us optimize the interface without giving developpers
> (not users!) strong garanties.
>
> Maybe the ALPN and protocol selection hooks could be marked as such
> too, we could have the need to change a RUN_ONCE to a RUN_ALL for
> those and shouldn't give too much garanties to other modules.
>
> So if we can find a good wording to also express "prod-ready" (I guess
> it already runs successfully in some prods), I'm +1.


Not to be a wet blanket, but I don't think we have any real data for
the production part of that. We could say the interfaces,
configuration and defaults are experimental in the volatile sense.

Re: mod_h2 CTR (Was: Re: svn commit: r1705257 - /httpd/httpd/trunk/modules/http2/config.m4)

Posted by Yann Ylavic <yl...@gmail.com>.
On Fri, Sep 25, 2015 at 4:31 PM, Jim Jagielski <ji...@jagunet.com> wrote:
> I'd like to propose that we consider mod_h2 in 2.4.x as CTR and
> in the same category as mod_lua... That is, somewhat "prod-experimental"
> in that some features may still change, but stable enough to be
> usable in production environs.

I like "prod-experimental" because I think it is production ready, and
still it lets us optimize the interface without giving developpers
(not users!) strong garanties.

Maybe the ALPN and protocol selection hooks could be marked as such
too, we could have the need to change a RUN_ONCE to a RUN_ALL for
those and shouldn't give too much garanties to other modules.

So if we can find a good wording to also express "prod-ready" (I guess
it already runs successfully in some prods), I'm +1.