Hi Conrad,
On Tue, May 20, 2014 at 02:46:15PM +0200, Conrad Hoffmann wrote:
> Hey Willy,
>
> > That's what I initially thought, because I didn't remember there was this
> > common fork_poller() function, but what you did is better, each part does
> > its own job, that's really clean this way.
> >
Hey Willy,
> That's what I initially thought, because I didn't remember there was this
> common fork_poller() function, but what you did is better, each part does
> its own job, that's really clean this way.
>
> Can you please provide a commit message summarizing the issue you faced
> and the sol
Hi Conrad,
On Mon, May 19, 2014 at 07:26:54PM +0200, Conrad Hoffmann wrote:
> >> Works all the time for me. Running lsof before and after the signal
> >> neatly shows how the worker closed the listening socket but the parent
> >> still has it.
> >
> > I'm just confused with what the "parent" here
Hey Willy,
On 05/19/2014 06:30 AM, Willy Tarreau wrote:
>> Here is a nice way to trigger it:
>> - run haproxy in deamon mode, but with nbproc 1
>> - open a HTTP keep-alive connection (and keep it open)
>> - send USR1 to the only worker (it should stay running, because of the
>>ongoing conne
Hi Conrad,
On Sat, May 17, 2014 at 08:31:46PM +0200, Conrad Hoffmann wrote:
> Hi Willy,
>
> On 05/17/2014 04:39 AM, Willy Tarreau wrote:
> > On 05/15/2014 11:26 AM, Conrad Hoffmann wrote:
> >> If haproxy has been running for a certain amount of time (this part
> >> makes reproduction cumbersome),
Hi Willy,
On 05/17/2014 04:39 AM, Willy Tarreau wrote:
> On 05/15/2014 11:26 AM, Conrad Hoffmann wrote:
>> If haproxy has been running for a certain amount of time (this part
>> makes reproduction cumbersome), we quite often run into the following
>> issue when restarting haproxy: if some of the w
Hi Conrad,
first, thanks for your detailed analysis.
[ merging your two mails for an easier response ]
On 05/15/2014 11:26 AM, Conrad Hoffmann wrote:
> If haproxy has been running for a certain amount of time (this part
> makes reproduction cumbersome), we quite often run into the following
> is
Hi,
I just figured out some of the missing pieces. When sending USR1 to the
workers, they just close() their listening sockets instead of shutdown()
(like they do for TTOU), causing the file descriptor to stay open in the
parent process, which is why there are still epoll events coming in for it.
Hi everyone,
I am still somewhat new to haproxy, so I maybe missing a few bits and
pieces here. If so, don't hesitate to educate me :)
First of all, we are using HAProxy here at SoundCloud, so a big thanks
to everyone who invested time in this wonderful project!
We are very keen on using SSL ter
9 matches
Mail list logo
