Ruby Forum NGINX > module proposal: BOSH

Posted by Adam Michaels (Guest)
on 07.08.2008 06:41
(Received via mailing list)
After much headache with Ejabberd and some internal discussions, we
talked about Nginx supporting the BOSH protocol. This would allow simple
creation of scalable instant message systems and anything else that
requires stateless connections to act statefully.

http://en.wikipedia.org/wiki/BOSH

Nginx would hold 2 client connections and create a persistent connection
to upstream servers. To even begin to make this possible, Nginx would
need to support persistent connections to upstreams. What is the status
of that feature? Its been listed on the feature request page in the wiki
for a while.

Anyone interested in seeing this?
Posted by alex clemesha (Guest)
on 07.08.2008 06:50
(Received via mailing list)
On Wed, Aug 6, 2008 at 9:36 PM, Adam Michaels 
<adam@digitalagemedia.net>wrote:

> feature? Its been listed on the feature request page in the wiki for a
> while.
>
> Anyone interested in seeing this?


yes!

-alex
Posted by mike (Guest)
on 07.08.2008 08:00
(Received via mailing list)
On 8/6/08, Adam Michaels <adam@digitalagemedia.net> wrote:

> Anyone interested in seeing this?

i don't think i'd be using it for quite some time but it does sound
cool and yet another thing nginx can excel in.
Posted by Manlio Perillo (Guest)
on 07.08.2008 08:49
(Received via mailing list)
Adam Michaels ha scritto:
> After much headache with Ejabberd 

Can you give more detail?
Thanks.


 > [...]


Manlio Perillo
Posted by Igor Clark (Guest)
on 07.08.2008 11:40
(Received via mailing list)
I can obviously see how clever a lot of it is, but am I the only
person who finds all this Comet/Bayeux/BOSH/Javascript "TCP sockets"/
etc stuff slightly unsettling? If you push and bend (e.g.) HTTP to do
all this stuff it was never intended to do, won't it, and all the
infrastructure supporting it, eventually just break? OK, so there's
Jetty and nginx can handle loads of keepalives and so on and so forth,
but ...

If people want bidirectional comms with browsers, if they want publish/
subscribe mechanisms, wouldn't all this work by very clever people be
better directed towards getting a sensible security infrastructure in
place to allow listening sockets in client sandboxes, rather than
building ever bigger wrappers with sticky tape and bits of string?

</devil's advocate>
Posted by Gabriel Ramuglia (Guest)
on 07.08.2008 18:59
(Received via mailing list)
A listening socket on the client isn't going to really work because of
firewalls and other network nastiness. And HTTP is the preferred
method for this kind of thing because it works on browsers without
modifying them. Assuming a new standard were made, it would be a good
number of years before the browsers that most people use would support
them, so the standard would be useless anyway, whereas http works now.
Posted by rkmr.em@gmail.com (Guest)
on 07.08.2008 19:00
(Received via mailing list)
On Wed, Aug 6, 2008 at 9:36 PM, Adam Michaels 
<adam@digitalagemedia.net>wrote:

> feature? Its been listed on the feature request page in the wiki for a
> while.
>
> Anyone interested in seeing this?
>
yes, this would be super awesome! +1
Posted by Du Song (Guest)
on 11.08.2008 10:13
(Received via mailing list)
Awesome if have!

Regards,
Du Song
Posted by Spil Games (spilgames)
on 12.08.2008 15:53
Adam Michaels wrote:
> Nginx would need to support persistent connections to upstreams. What is the
> status of that feature? Its been listed on the feature request page in the
> wiki for a while.

I would be very interested to see support for persistent connections to 
upstreams, and the company I'm working for would be willing to 
(co)sponsor such a feature. I already contacted Evan Miller about this, 
but he doesn't have enough time to pick this up currently. Anyone else 
capable of and willing to pick this up?
Posted by Kon Wilms (Guest)
on 12.08.2008 16:08
(Received via mailing list)
On Tue, Aug 12, 2008 at 6:53 AM, Spil Games <lists@ruby-forum.com> 
wrote:
> Adam Michaels wrote:
>> Nginx would need to support persistent connections to upstreams. What is the
>> status of that feature? Its been listed on the feature request page in the
>> wiki for a while.
>
> I would be very interested to see support for persistent connections to
> upstreams, and the company I'm working for would be willing to
> (co)sponsor such a feature. I already contacted Evan Miller about this,
> but he doesn't have enough time to pick this up currently. Anyone else
> capable of and willing to pick this up?

My company is also willing to add to the bounty for persistent
connection support to upstreams, notably memcached. I simply do not
have the time.

Cheers
Kon
Posted by Phillip B Oldham (Guest)
on 09.10.2008 10:57
Attachment: phill.vcf (262 Bytes)
(Received via mailing list)
Adam Michaels wrote:
> is the status of that feature? Its been listed on the feature request 
> page in the wiki for a while.
>
> Anyone interested in seeing this?
Has any progress been made with this?

--

*Phillip B Oldham*
The Activity People
phill@theactivitypeople.co.uk <mailto:phill@theactivitypeople.co.uk>

------------------------------------------------------------------------

*Policies*

This e-mail and its attachments are intended for the above named
recipient(s) only and may be confidential. If they have come to you in
error, please reply to this e-mail and highlight the error. No action
should be taken regarding content, nor must you copy or show them to 
anyone.

This e-mail has been created in the knowledge that Internet e-mail is
not a 100% secure communications medium, and we have taken steps to
ensure that this e-mail and attachments are free from any virus. We must
advise that in keeping with good computing practice the recipient should
ensure they are completely virus free, and that you understand and
observe the lack of security when e-mailing us.
Posted by Spil Games (spilgames)
on 09.10.2008 11:23
Phillip B Oldham wrote:
> Has any progress been made with this?


AFAIK *someone* still has to pickup the persistent connection thing. 
Several companies are willing to pay bounties for this, and I asked Evan 
Miller to do this, but he is currently out of OSS time. I still have 
this on my radar though and am still actively seeking for a developer 
willing to pick this up. No luck yet though.

Anyone else made any progress?
Posted by Maxim Dounin (Guest)
on 09.10.2008 12:25
(Received via mailing list)
Hello!

On Thu, Oct 09, 2008 at 11:23:18AM +0200, Spil Games wrote:

>Anyone else made any progress?
I've recently posted here a patch that will allow nginx to know
when request is completed even if backend server wasn't closed
connection.  The patch is currently waiting for Igor's review.

There are two things left for http persistent connections support:
chunked encoding (should be relatively easy with the mentioned
patch) and saving connections for later use instead of closing
them (relatively easy as is).

Maxim Dounin
Posted by Spil Games (spilgames)
on 09.10.2008 15:42
Maxim Dounin wrote:
> There are two things left for http persistent connections support:
> chunked encoding (should be relatively easy with the mentioned
> patch) and saving connections for later use instead of closing
> them (relatively easy as is).

No interest in collecting the bounty?
Posted by Maxim Dounin (Guest)
on 09.10.2008 23:59
(Received via mailing list)
Hello!

On Thu, Oct 09, 2008 at 03:42:25PM +0200, Spil Games wrote:

>Maxim Dounin wrote:
>> There are two things left for http persistent connections support:
>> chunked encoding (should be relatively easy with the mentioned
>> patch) and saving connections for later use instead of closing
>> them (relatively easy as is).
>
>No interest in collecting the bounty?

Feel free to donate if/when I'll do it. :)

Maxim Dounin