[solved] Cpanel + LSAPI 6.0 + LSWS 4.2 Anomaly

[DraCoola]

is it true that LSAPI 6.0 using much more swap than LSAPI 5.5?
the left one is still using LSAPI 5.5 :

Uploaded with ImageShack.us

after using LSAPI 6.0 I have to raise up externall app >> lsphp5 >> memory soft/hard limit to at least 505 Mb to avoid randomly "Fatal error: Out of memory"
change from "Run On Start Up" to "suEXEC daemon" are the same on result on memory cost.



and for lsws 4.2, it cannot compile php 5.3.x through WHM while using it :

Uploaded with ImageShack.us

so it force me to use LSWS 4.1.13 with LSAPI 6.0 until now.
is there anyway I can recompile php using the old LSAPI 5.5?

 

[DraCoola]

--pardon me, double post---

 

[NiteWave]

is there anyway I can recompile php using the old LSAPI 5.5

try:modify
/usr/local/lsws/admin/html/utility/build_php/buildconf.inc.php
define ('LSAPI_VERSION', '6.0');
to
define ('LSAPI_VERSION', '5.5');

 

[DraCoola]

Thanks Nitewave.
Unfortunatelly I cannot found define ('LSAPI_VERSION', '6.0'); on that file :

<?php

define ('LEGAL', 1);
define ('TITLE', 'Compile PHP with LSAPI');
define ('OPTION_VERSION', 2);

define ('BUILD_DIR', $_SERVER['LS_SERVER_ROOT'] . 'phpbuild');
define ('LAST_CONF', BUILD_DIR . '/savedconfig.'); // actual file will include . php base version.

define ('DEFAULT_INSTALL_DIR', $_SERVER['LS_SERVER_ROOT'].'lsphp'); // actual dir will include . php base version.

$PHP_VER = array('5'=>
                 array(
                         '5.4.6',
                         '5.4.5',
                         '5.3.16',
                         '5.3.15',
                         '5.3.8',
                         '5.3.6',
                         '5.2.17',
                     '5.2.9',
                     '5.1.6'),
                 '4'=>
                 array('4.4.9',
                       '4.4.8'));

define ('LSAPI_VERSION', '5.5');
define ('SUHOSIN_VERSION', '0.9.33');
define ('APC_VERSION', '3.1.9');
define ('XCACHE_VERSION', '2.0.1');
define ('MEMCACHE_VERSION', '2.2.6');
//define ('MEMCACHED_VERSION', '1.0.2');



$DEFAULT_PHP_PARAMS = array(
        '5' => '--with-mysqli --with-zlib --with-gd --enable-shmop --enable-track-vars --enable-sockets --enable-sysvsem --enable-sysvshm --enable-magic-quotes --enable-mbstring --with-iconv',
        '4' => '--with-mysql  --with-zlib --with-gd --enable-shmop --enable-track-vars --enable-sockets --enable-sysvsem --enable-sysvshm --enable-magic-quotes --enable-mbstring');


include_once( 'buildfunc.inc.php' );

?>

But my phpinfo() is showing for Server API >> LiteSpeed V6.0

 

[webizen]

follow this (http://www.litespeedtech.com/suppor...=matching php#step_2build_matching_php_binary) to build matching php in lsws admin console. it should get you lsapi 5.5 in 4.1.13.

 

[mistwang]

I do not think there is any change related to memory usage between lsapi 5.5 and 6.0 .
you have to test it on the same server with same PHP release.

to get back to 5.5, you can manually download php-litespeed package and over write *.c and *.h under php-xxxx/sapi/litespeed, the run make, copy over php-xxxx/sapi/php to /usr/local/lsws/fcgi-bin/lsphp-5.xxx, you can keep both 5.5 and 6.0 binary, and switch between them with changing symbolic link of lsphp5.

 

[DraCoola]

Thank you George.
But i cannot run make after overwrite *.c and *.h :

root@rooted [/home/cpeasyapache/ls.src/php-5.2.17/sapi/litespeed]# make
make: *** No targets specified and no makefile found.  Stop.

 

[mistwang]

run make under php-xxxx/ directory.

 

[DraCoola]

thanks George, I will try it soon.

by the way using LSWS 4.2 -suEXEC daemon- new feature makes php login session suddenly throw out.
I have got two complaints regarding to this.
and as my 1st post, it using swap much even phisycal RAM space have enough space.

I don't know what I am missing, but my two boxes bring the same behaviour about login session and force using swap.

 

[webizen]

...

I don't know what I am missing, but my two boxes bring the same behaviour about login session and force using swap.

Can you elaborate the issue?

Also for the swap usage, it means at one point that much swap was used (when there was no free memory). It does not mean your server is currently using swap while free memory still available.

 

[DraCoola]

pardon me for my long delay of reply.

yesterday I am using lsapi 6.0 again, all issue wroted on this thread seems gone except one :
all hosting accounts using PHPIniDir (custom php.ini) are hard to login to their site (as reposted : wordpress, prestashop).
or if they can finally logged in but will throw out at aproximatelly 3 minutes or at random duration.

 

[webizen]

disable suEXEC daemon mode to work around this.

there is an issue with per user dir php.ini in suEXEC daemon mode due to php internal handling. hence what you experienced.

in 4.2.1 (beta), LSWS automatically disables suEXEC daemon mode on per user dir php.ini. you can upgrade by issuing '/usr/local/lsws/admin/misc/lsup.sh -v 4.2.1'

 

[DraCoola]

all of issues that i have been told in this thread are fixed now by lsws 4.2.1 stable release and lsapi 6.1

thank you so very much

 

[mistwang]

there is a new update of 4.2.1 release, please upgrade
/usr/local/lsws/admin/misc/lsup.sh -f -v 4.2.1

 

[DraCoola]

there is a new update of 4.2.1 release, please upgrade
/usr/local/lsws/admin/misc/lsup.sh -f -v 4.2.1

is that new 4.2.1 release more newest than the new 4.2.1 from yesterday?
because the old 4.2.1 makes me still have to raise up Memory Soft Limit of External App >> lsphp5 to at least 650M

bellow of 650M value will send many emails of "LiteSpeed Web server restarts automatically to fix 503 Errors"
and also cpu load average is getting higher than usual.

 

[mistwang]

If you still experiencing this problem
http://www.litespeedtech.com/support/forum/showthread.php?t=6412

should upgrade again. It may not have any effect on memory limit though.

 

[DraCoola]

It may not have any effect on memory limit though.

i have experiencing that with cloudlinux boxes only.
is that possible if 503 page which generated by cloudlinux will automatically restarting by lsws (Auto Fix 503 Error)?

because so many "restarts automatically to fix 503 Errors" emails came from lsws with clouddlinux until now.

 

[Alfa-Host]

i have experiencing that with cloudlinux boxes only.
is that possible if 503 page which generated by cloudlinux will automatically restarting by lsws (Auto Fix 503 Error)?

because so many "restarts automatically to fix 503 Errors" emails came from lsws with clouddlinux until now.

I have the same problem + my server crash some hour after enable LVE or cagefs from litespeed and i must HW reboot server !

DraCoola, do you have crash problem ? if no , what is your kernel version ? (uname -a)

 

[DraCoola]

I have the same problem + my server crash some hour after enable LVE or cagefs from litespeed and i must HW reboot server !

DraCoola, do you have crash problem ? if no , what is your kernel version ? (uname -a)

hi there Alfa-Host, my current kernel is :

2.6.18-408.8.2.el5.lve0.8.61.3 #1 SMP Wed Jul 11 06:49:35 EDT 2012 x86_64 x86_64 x86_64 GNU/Linux

I didn't experience any crash which makes me to do hard reboot.
I just got 503 error emails which told restarting lsws many times per day for everyday.
but before using of these newest lsws and lsapi, 503 error emails wasn't arrive at all.

 

[mistwang]

i have experiencing that with cloudlinux boxes only.
is that possible if 503 page which generated by cloudlinux will automatically restarting by lsws (Auto Fix 503 Error)?

because so many "restarts automatically to fix 503 Errors" emails came from lsws with clouddlinux until now.

Can you switch back to 4.1.13 and see if you get 503 errors as well?
I have noticed on one customer's server using cloudlinux, that lsphp5 processes was killed by signal 9, which causes 503 errors, not sure who send signal 9. I wonder if it is caused by cloudlinux update or not.