Everything
php
Learn Community Releases The PHP Project
Home » Project » Newsgroups » php.internals » Upgrading www.php.net to PHP 5.0 »

Upgrading www.php.net to PHP 5.0

php.internals

Andi Gutmans

22 years ago
Hi, As I understand that www.php.net (the main web site) works fine with PHP 5, can someone with root access upgrade to PHP 5 on www.php.net? As discussed here, it'd be good to make this switch in preparation for the upcoming release. FYI, I'm hopping on a flight now so I won't be reading email in the next 36 hours. Thanks, Andi

Wez Furlong

22 years ago
If someone does upgrade it, please copy the getloadavg() function from the sources in ~wez. If no one has done it by the time I get back (I'm also plane hopping), I'll do it. --Wez. ----- Original Message ----- From: "Andi Gutmans" <andi@zend.com> To: <internals@lists.php.net> Sent: Sunday, June 27, 2004 8:28 AM Subject: [PHP-DEV] Upgrading www.php.net to PHP 5.0
> Hi, > > As I understand that www.php.net (the main web site) works fine with PHP
5,
> can someone with root access upgrade to PHP 5 on www.php.net? > As discussed here, it'd be good to make this switch in preparation for the > upcoming release. > FYI, I'm hopping on a flight now so I won't be reading email in the next
36

Derick Rethans

22 years ago
On Sun, 27 Jun 2004, Wez Furlong wrote:
> If someone does upgrade it, please copy the getloadavg() function from the > sources in ~wez. > > If no one has done it by the time I get back (I'm also plane hopping), I'll > do it.
Do not upgrade it: 10:50 <@edink> i had to fix ez1 over the weekend 10:50 <@Derick> now bugs. 10:50 <@Derick> ah, you already did 10:50 <@Derick> what was broken? 10:51 <@edink> php5 10:51 <@Derick> hehe 10:51 <@edink> leaks like hell 10:51 <@Derick> ouch 10:51 <@edink> all mem was used up 10:51 <@edink> so a put maxreqperchild to 20 10:51 <@Derick> yikes 10:52 <@edink> some httpd processes were 600MB ;) regards, Derick

Stanislav Malyshev

22 years ago
DR>>10:51 <@edink> leaks like hell DR>>10:51 <@Derick> ouch DR>>10:51 <@edink> all mem was used up DR>>10:51 <@edink> so a put maxreqperchild to 20 DR>>10:51 <@Derick> yikes DR>>10:52 <@edink> some httpd processes were 600MB ;) I suggest to compile it in debug mode and see where the leaks are...
-- Stanislav Malyshev, Zend Products Engineer stas@zend.com http://www.zend.com/ +972-3-6139665 ext.115

Zeev Suraski

22 years ago
Edin, Can you try disabling ZEND_MM and see if the problem persists? (it's in Zend/zend_mm.h, #undef ZEND_MM). Zeev At 11:54 28/06/2004, Derick Rethans wrote:

Derick Rethans

22 years ago
On Mon, 28 Jun 2004, Zeev Suraski wrote:
> Edin, > > Can you try disabling ZEND_MM and see if the problem persists? (it's in > Zend/zend_mm.h, #undef ZEND_MM).
Will do. Derick

Sterling Hughes

22 years ago
yep, this is what I found too, and have been saying for nearly a year: http://www.edwardbear.org/blog/archives/000129.html#000129 maybe one of these days we'll give up on the ZEND_MM, which doesn't by us much anyway and has leaked since well before and after beta 1. Ok, that's my annoying I told you so for the month, back to lurking :) -sterling On Mon, 28 Jun 2004 13:07:38 +0200 (CEST), Derick Rethans <derick@php.net> wrote:

Andi Gutmans

22 years ago
Sterling, As far as I can tell, it's not ZEND_MM which is leaking, but after I fixed a bug in zend_alloc.c, these problems started happening. We are using the same zend_alloc.c code for ZEND_MM and allocator on Win32, so in any case, this problem needs to be fixed. It might be that it happens when ZEND_MM is enabled but it doesn't mean that the fault is in ZEND_MM. Check your facts first or even better, why don't you help debug the problem? Andi At 08:46 AM 6/28/2004 -0700, Sterling Hughes wrote:

Sterling Hughes

22 years ago
well, disabling ZEND_MM does *fix* the/a leak. It was disabled when we released beta 1 and has been subsequently re-enabled. As for checking my facts and helping to debug the problem, I'm the one who found the leak and who found the "fix" (a change which prevents the problem from occuring) nearly a year ago. I stopped debugging it around the time I got tired of doing real work on PHP, but your point is taken, as I've told qa people a thousand times, its easy to always be right when you haven't written any code. -sterling On Mon, 28 Jun 2004 08:50:15 -0700, Andi Gutmans <andi@zend.com> wrote:

Andi Gutmans

22 years ago
BTW, besides ZEND_MM being slightly faster on a few benchmarks I made (although it's platform dependant so likely malloc could be faster on others), I don't mind disabling it for 5.0.0. That said, there's a bug lurking which we need to find before, and as I mentioned, I don't think it's in the ZEND_MM code. Andi At 08:50 AM 6/28/2004 -0700, Andi Gutmans wrote: