Hi Slava,
first of all, your patch does handle only the conn_pool timeouts and this is
not enough. Just as example: open tcp/ip conn to server but server doesn't
send any responses (overloaded or any other reason). In this case we have ,
possible all connections that are not winthin conn_pool but should be
timeouted...
The second point is, just look in your code:
static void conn_pool_clean(void) {
...
mutex_lock(conn_pool_lock);
^^^^^
that means , we can't use pool as long cleanup running... So I just don't see
any advantages to using threaded version.
...
}
so I'm -1 for (1)* and +1 for (2).
* as long none benchmarks shows that (1) really faster
P.S. David currently working on a better patch for (2) also just be patient...
On Wednesday 03 December 2003 08:02, Vjacheslav Chekushin wrote:
> I think that threaded version realy more clean and maintainable.
> As I wrote in previouse posts perfomance of non-threaded version
> depends on load, but for threaded version perfomance is constant.
>
> I am +1 for 1).
>
> Nisan Bloch wrote:
> > Hi
> >
> > I agree with Alexander here - if the threaded version shows no real
> > gains, then why include it - more source to maintain - more chance of
> > bugs etc. The non-thread ver is enough of a gain as is.
> >
> > +1 for 2
> >
> > Nisan
> >
> > At 12:31 PM 2003/12/02, Alexander Malysh wrote:
> >> Hi,
> >>
> >> On Tuesday 02 December 2003 11:10, Stipe Tolj wrote:
> >> > David Schmitz wrote:
> >> > > attached are two patches for the HTTP-timeout thing. The first one
> >>
> >> uses
> >>
> >> > > a threaded-approach and the second one works without threads. As
> >> > > some tests have shown, the threaded version only performs better in
> >>
> >> high-load
> >>
> >> > > situations, and even then only marginally. Thus, I would like you to
> >> > > test vote on one of four possibilities:
> >> > > 1. commit thread-solution
> >> > > 2. commit non-thread-solution
> >> > > 3. merge both and make it selectable with a precompiler-directive
> >> > > 4. delete the whole thing.
> >> >
> >> > if we'd pick optionm 3, how long would it take?
> >>
> >> Stipe, are you really want more complicated http-module as needed?
> >> I would propose to stress-test/benchmark 2 patches and vote for only
> >> one for
> >> inclusion...
> >>
> >> P.S. I'm +1 for 4) ;) (just a joke)
> >>
> >> > Stipe
> >> >
> >> > mailto:[EMAIL PROTECTED]
> >> > -------------------------------------------------------------------
> >> > Wapme Systems AG
> >> >
> >> > M�nsterstr. 248
> >> > 40470 D�sseldorf, NRW, Germany
> >> >
> >> > phone: +49.211.74845.0
> >> > fax: +49.211.74845.299
> >> >
> >> > mailto:[EMAIL PROTECTED]
> >> > http://www.wapme-systems.de/
> >> > -------------------------------------------------------------------
> >> >
> >> > -----BEGIN PGP PUBLIC KEY BLOCK-----
> >> > Version: GnuPG v1.2.2 (Cygwin)
> >> >
> >> > mIsEP6mcYwEEAMDnUiUwrbb+xwTFWN6TxF2+XZu7/alwJMeCwMBRvXtPZqfjpPhS
> >> > OkBpU0F4TrVuugz1HINTSaJTYq10AzDQXp5NkyWgckqW79nPAWuOX0dicbJk+cN2
> >> > nM2TI4KaxUDe6u8hghNEnH/i2lXsUu9apnP/iixzV81VC2je3uc9hZpnAAYptEVT
> >> > dGlwZSBUb2xqIChUZWNobm9sb2d5IENlbnRlciAmIFJlc2VhcmNoIExhYikgPHRv
> >> > bGpAd2FwbWUtc3lzdGVtcy5kZT6ItAQTAQIAHgUCP6mcYwIbAwYLCQgHAwIDFQID
> >> > AxYCAQIeAQIXgAAKCRABV0w1BqPYRuSqA/wPzsQxao2YePENCtgRTrO86U6zg3sl
> >> > OcS6CJFI4FZP5h/xD3GRsNH1+MPSvZlomDdpFnr547DGz/Kq9MXuQwVvlVig5yWZ
> >> > K5dtKp1r5YLhxJQBhfirZbRFFnYmf19f18J8OoS28tuFVftDl1AIwJS3HLyBTv6H
> >> > g2HyLAEKQIp30Q==
> >> > =aYCI
> >> > -----END PGP PUBLIC KEY BLOCK-----
> >>
> >> --
> >> Best regards / Mit besten Gr��en aus D�sseldorf
> >>
> >> Dipl.-Ing.
> >> Alexander Malysh
> >> ___________________________________________
> >>
> >> Centrium GmbH
> >> Vogelsanger Weg 80
> >> 40470 D�sseldorf
> >>
> >> Fon: +49 (0211) 74 84 51 80
> >> Fax: +49 (0211) 277 49 109
> >>
> >> email: [EMAIL PROTECTED]
> >> web: www.centrium.de
> >> msn: [EMAIL PROTECTED]
> >> icq: 98063111
> >> ___________________________________________
> >>
> >> Please avoid sending me Word, Excel or PowerPoint attachments.
> >> See http://www.fsf.org/philosophy/no-word-attachments.html
--
Best regards / Mit besten Gr��en aus D�sseldorf
Dipl.-Ing.
Alexander Malysh
___________________________________________
Centrium GmbH
Vogelsanger Weg 80
40470 D�sseldorf
Fon: +49 (0211) 74 84 51 80
Fax: +49 (0211) 277 49 109
email: [EMAIL PROTECTED]
web: www.centrium.de
msn: [EMAIL PROTECTED]
icq: 98063111
___________________________________________
Please avoid sending me Word, Excel or PowerPoint attachments.
See http://www.fsf.org/philosophy/no-word-attachments.html
