Buy commercial curl support from WolfSSL. We help you work
out your issues, debug your libcurl applications, use the API, port to new
platforms, add new features and more. With a team lead by the curl founder
himself.
Re: HTTPS traffic much slower after upgrade from 7.85.0 to 7.87.0
- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]
From: Fabian Keil via curl-library <curl-library_at_lists.haxx.se>
Date: Thu, 9 Feb 2023 12:00:36 +0100
Daniel Stenberg via curl-library <curl-library_at_lists.haxx.se> wrote on 2023-02-09 at 11:26:09:
> On Thu, 9 Feb 2023, Stephan Mühlstrasser via curl-library wrote:
>
> > I'm attempting to upgrade libcurl in an application from 7.85.0 to 7.87.0.
> > When I run tests against an internal test server, I see a major slowdown for
> > HTTPS connections. It is worst when proxies are involved, but I see also a
> > slowdown without proxies.
>
> Yes, we are already aware of this regression and several fixes have been
> landed in git to address this. Actually to the level that in our tests, git
> master is much faster to first byte received than ever before (as far as I
> have verified in my tests at least).
>
> It would be helpful if you could verify that the slowdown is gone for you as
> well if you build and test with current git master.
>
> This regression happened because we don't do regular performance tests, and as
> everything still kept working, albeit too slowly, we did not notice this in
> time before the release! :-O
In the Privoxy project we currently don't do regular and
systematic performance tests either ...
While I started to collect some TLS benchmarks [0] a while
ago I haven't updated them for quite some time.
There recently was a significant performance regression [1]
that was fixed in time [2] but delayed the 3.0.34 release.
As some of you know, there are already patches available
to use a subset of the curl tests with Privoxy [3] but
due to the lack of curl performance tests Privoxy's
performance regression wasn't detected with automated
testing.
Long story short: I'd be interested to collaborate
on creating curl performance tests so I can leverage
them to test Privoxy as well.
Fabian
[0]: <https://www.fabiankeil.de/gehacktes/privoxy-tls-benchmarks/>
[1]: <https://lists.privoxy.org/pipermail/privoxy-devel/2023-January/000579.html>
[2]: <https://www.privoxy.org/gitweb/?p=privoxy.git;a=commitdiff;h=18eeee66a85d364b56>
[3]: <https://www.fabiankeil.de/sourcecode/privoxy/privoxy-cts-tests-2022-06-03.diff>
Received on 2023-02-09
Date: Thu, 9 Feb 2023 12:00:36 +0100
Daniel Stenberg via curl-library <curl-library_at_lists.haxx.se> wrote on 2023-02-09 at 11:26:09:
> On Thu, 9 Feb 2023, Stephan Mühlstrasser via curl-library wrote:
>
> > I'm attempting to upgrade libcurl in an application from 7.85.0 to 7.87.0.
> > When I run tests against an internal test server, I see a major slowdown for
> > HTTPS connections. It is worst when proxies are involved, but I see also a
> > slowdown without proxies.
>
> Yes, we are already aware of this regression and several fixes have been
> landed in git to address this. Actually to the level that in our tests, git
> master is much faster to first byte received than ever before (as far as I
> have verified in my tests at least).
>
> It would be helpful if you could verify that the slowdown is gone for you as
> well if you build and test with current git master.
>
> This regression happened because we don't do regular performance tests, and as
> everything still kept working, albeit too slowly, we did not notice this in
> time before the release! :-O
In the Privoxy project we currently don't do regular and
systematic performance tests either ...
While I started to collect some TLS benchmarks [0] a while
ago I haven't updated them for quite some time.
There recently was a significant performance regression [1]
that was fixed in time [2] but delayed the 3.0.34 release.
As some of you know, there are already patches available
to use a subset of the curl tests with Privoxy [3] but
due to the lack of curl performance tests Privoxy's
performance regression wasn't detected with automated
testing.
Long story short: I'd be interested to collaborate
on creating curl performance tests so I can leverage
them to test Privoxy as well.
Fabian
[0]: <https://www.fabiankeil.de/gehacktes/privoxy-tls-benchmarks/>
[1]: <https://lists.privoxy.org/pipermail/privoxy-devel/2023-January/000579.html>
[2]: <https://www.privoxy.org/gitweb/?p=privoxy.git;a=commitdiff;h=18eeee66a85d364b56>
[3]: <https://www.fabiankeil.de/sourcecode/privoxy/privoxy-cts-tests-2022-06-03.diff>
-- Unsubscribe: https://lists.haxx.se/listinfo/curl-library Etiquette: https://curl.se/mail/etiquette.html
- application/pgp-signature attachment: OpenPGP digital signature