Re: [PHP-DEV] Re: PHP-FPM state

2014-11-20 Thread Remi Collet
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Le 20/11/2014 10:34, Remi Collet a écrit : > P.S. Still TODO: listen_allowed_clients is now aware of IPv4/IPv6, > but it could make sense to make it aware of hostname (using > getaddrinfo), and set all found addresses as allowed. Probably not > urgent

Re: [PHP-DEV] Re: PHP-FPM state

2014-11-20 Thread Remi Collet
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 I think the situation is now better. At least test suite grow from 3 to 15 tests (+1 for Apparmor) and all recent bugs / changes are covered. Notice: test suite is 5.6+ only as rely on --force-stderr Some minor changes are 5.6+ only 09f9f7ce503097

Re: [PHP-DEV] Re: PHP-FPM state

2014-11-17 Thread Ferenc Kovacs
On Mon, Nov 17, 2014 at 4:24 PM, Remi Collet wrote: > -BEGIN PGP SIGNED MESSAGE- > Hash: SHA1 > > Le 15/11/2014 11:11, Ferenc Kovacs a écrit : > > > Yeah, the test coverage of fpm is lacking, and while this PR > > contained a test for testing the happy path (that you can listen on > > ipv

Re: [PHP-DEV] Re: PHP-FPM state

2014-11-17 Thread Remi Collet
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Le 15/11/2014 11:11, Ferenc Kovacs a écrit : > Yeah, the test coverage of fpm is lacking, and while this PR > contained a test for testing the happy path (that you can listen on > ipv6) but there were no tests which could have helped us to spot > the

Re: [PHP-DEV] Re: PHP-FPM state

2014-11-17 Thread Ferenc Kovacs
On Mon, Nov 17, 2014 at 12:38 PM, Pierre Joye wrote: > > On Nov 17, 2014 4:26 PM, "Matteo Beccati" wrote: > > > > On 15/11/2014 11:11, Ferenc Kovacs wrote: > > > 2014.11.15. 8:28 ezt írta ("Remi Collet" > > > ): > > >> > > > FPM state in 5.5.19 / 5.6.3 is quite bad :( > > > > Maybe I'm saying so

Re: [PHP-DEV] Re: PHP-FPM state

2014-11-17 Thread Pierre Joye
On Nov 17, 2014 4:26 PM, "Matteo Beccati" wrote: > > On 15/11/2014 11:11, Ferenc Kovacs wrote: > > 2014.11.15. 8:28 ezt írta ("Remi Collet" > > ): > >> > > FPM state in 5.5.19 / 5.6.3 is quite bad :( > > Maybe I'm saying something unpopular, but wouldn't it be better to > rollback the fpm changes,

Re: [PHP-DEV] Re: PHP-FPM state

2014-11-17 Thread Matteo Beccati
On 15/11/2014 11:11, Ferenc Kovacs wrote: > 2014.11.15. 8:28 ezt írta ("Remi Collet" > ): >> > FPM state in 5.5.19 / 5.6.3 is quite bad :( Maybe I'm saying something unpopular, but wouldn't it be better to rollback the fpm changes, release 5.5.20 and 5.6.4 and keep fpm+ipv6 for master? Cheers -

Re: [PHP-DEV] Re: PHP-FPM state

2014-11-17 Thread Remi Collet
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Le 15/11/2014 08:27, Remi Collet a écrit : I have also fixed: https://bugs.php.net/68428 listen.allowed_clients is IPv4 only > SO PLEASE, everyone who can test these patches, please test and > give feedback [1] Really lot of change

Re: [PHP-DEV] Re: PHP-FPM state

2014-11-15 Thread Ferenc Kovacs
2014.11.15. 8:28 ezt írta ("Remi Collet" ): > > -BEGIN PGP SIGNED MESSAGE- > Hash: SHA1 > > FPM state in 5.5.19 / 5.6.3 is quite bad :( > > I just fix: > > https://bugs.php.net/68420 > listen=9000 listens to ipv6 localhost instead of all addresses > > This is obviously a reg

Re: [PHP-DEV] Re: PHP-FPM state

2014-11-14 Thread Remi Collet
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 FPM state in 5.5.19 / 5.6.3 is quite bad :( I just fix: https://bugs.php.net/68420 listen=9000 listens to ipv6 localhost instead of all addresses This is obviously a regression. Ok, workaround exists, using explicit address which see