Re: [fossil-users] Stress testing www4... (was Re: [Bug?] [server] Processes of Fossil popping up unexpectedly

2017-12-23 Thread Richard Hipp 
On 12/23/17, Olivier R.  wrote:
>
> Is it a problem if users only use Fossil 2.4 at home?

No.


> Does other users must rebuild on their side or is it enough to sync only?
>

No "fossil rebuild" is required.  Other users will not notice any difference.
-- 
D. Richard Hipp
d...@sqlite.org
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] Stress testing www4... (was Re: [Bug?] [server] Processes of Fossil popping up unexpectedly

2017-12-23 Thread Olivier R. 

Le 23/12/2017 à 02:14, Richard Hipp a écrit :


I did a half-baked idle timer in the latest check-in, using the
--max-latency option.

 fossil server --max-latency 30


I cloned the Fossil repo, built the new version and also relaunch Fossil 
with:


nohub fossil server --max-latency 30 &

Is it a problem if users only use Fossil 2.4 at home?
I assume it’s not, but I prefer to ask.

I also rebuilt the repo.
Does other users must rebuild on their side or is it enough to sync only?

Olivier
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] Stress testing www4... (was Re: [Bug?] [server] Processes of Fossil popping up unexpectedly

2017-12-22 Thread Richard Hipp 
On 12/22/17, John P. Rouillard  wrote:
>
> Well since reverse proxy servers use a client idle setting, would I be
> crazy to suggest the same for fossil?
>

I did a half-baked idle timer in the latest check-in, using the
--max-latency option.

fossil server --max-latency 30

The above will automatically terminate each request after 30 seconds,
finished or not.  I started out trying to implement a more complex
idle timeout, but that was a lot of code to test, and it is late here.
Maybe we can add something like that later.  Or maybe we can tell
people to use a reverse proxy if they are deploying in an environment
where a more advanced idle timeout is necessary to keep bad actors at
bay.

-- 
D. Richard Hipp
d...@sqlite.org
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Re: [fossil-users] Stress testing www4... (was Re: [Bug?] [server] Processes of Fossil popping up unexpectedly

2017-12-22 Thread John P. Rouillard 
Hi Richard:

In message

Re: [fossil-users] Stress testing www4... (was Re: [Bug?] [server] Processes of Fossil popping up unexpectedly

2017-12-22 Thread Richard Hipp 
On 12/22/17, John P. Rouillard  wrote:
>
> I am currently running 5
>
>   nc www4.fossil-scm.org 80
>
> processes and sending no input.

Ah hah!

I think the problem is this code:
http://www4.fossil-scm.org/artifact/9694a7dc?ln=1855-1858

That connection rate throttler is code that predates Fossil.  I copied
it from CVSTrac.  It is present in the very first check-in of Fossil
(2007-07-21).

MAX_PARALLEL is currently set to 2, which is a ridiculously low
number.  I have raised it to 50 on www4.fossil-scm.org and restarted
that server.  I should probably raise it to 1000 or so.  And instead
of sleeping, I'll just block until children die off.

This problem only comes up when running "fossil server" with direct
connections to the outside, and rogue outsiders starting up
connections and just sitting on them.  People using a reverse proxy
(ex: nginx) won't see the problem because the reverse proxy never sits
on a TCP/IP connection.

Give me a few minutes to come up with the best solution.  Suggestions
from the mailing list are also welcomed.

-- 
D. Richard Hipp
d...@sqlite.org
___
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users