Posts Tagged ‘client browser’

How to install and use memcached on Debian GNU / Linux to share php sessions between DNS round robined Apache webservers

Monday, November 9th, 2020

Reading Time: 8minutes

apache-load-balancing-keep-persistent-php-sessions-memcached-logo

Recently I had to come up with a solution to make A bunch of websites hosted on a machine to be high available. For the task haproxy is one of logical options to use. However as I didn't wanted to set new IP addresses and play around to build a cluster. I decided the much more simplistic approach to use 2 separate Machines each running Up-to-date same version of Apache Webserver as front end and using a shared data running on Master-to-Master MySQL replication database as a backend. For the load balancing itself I've used a simple 2 multiple DNS 'A' Active records, configured via the Bind DNS name server an Round Robin DNS load balancing for each of the domains, to make them point to the the 2 Internet IP addresses (XXX.XXX.XXX.4 and YYY.YYY.YYY.5) each configured on the 2 Linux servers eth0.

So far so good, this setup worked but immediately, I've run another issue as I found out the WordPress and Joomla based websites's PHP sessions are lost, as the connectivity by the remote client browser reaches one time on XXX…4 and one time on YYY…4 configured listerner on TCP port 80 and TCP p. 443. In other words if request comes up to Front end Apache worker webserver 1 with opened channel data is sent back to Client Browser and the next request is sent due to the other IP resolved by the DNS server to come to Apache worker webserver 2 of course webserver 2 has no idea about this previous session data and it gets confused and returns soemething like a 404 or 500 or any other error … not exciting really huh …

I've thought about work around and as I didn't wanted to involve thirty party stuff as Privoxy / Squid  / Varnish / Polipo etc. just as that would add extra complexity as if I choose to use haproxy from the beginning, after short investigation came to a reason to use memcached as a central PHP sessions storage.

php-memcached-apache-workers-webbrowser-keep-sessions-diagram
 

Why I choose memcached ?


Well it is relatively easy to configure, it doesn't come with mambo-jambo unreadable over-complicated configuration and the time to configure everything is really little as well as the configuration is much straight forward, plus I don't need to occupy more IP addresses and I don't need to do any changes to the already running 2 WebServers on 2 separate Linux hosts configured to be reachable from the Internet.
Of course using memcached is not a rock solid and not the best solution out there, as there is risk that if a memcached dies out for some reason all sessions stored in are lost as they're stored only in volatile memory, as well as there is a drawback that if a communication was done via one of the 2 webservers and one of them goes down sessions that were known by one of Apache's workers disappears.

So let me proceed and explain you the steps to take to configure memcached as a central session storage system.
 

1. Install memcached and php-memcached packages


To enable support for memcached besides installing memcached daemon, you need to have the php-memcached which will provide the memcached.so used by Apache loaded php script interpretter module.

On a Debian / Ubuntu and other deb based GNU / Linux it should be:

webserver1:~# apt-get install memcached php-memcached

TO use php-memcached I assume Apache and its support for PHP is already installed with lets say:
 

webserver1:~# apt-get install php libapache2-mod-php php-mcrypt


On CentOS / RHEL / Fedora Linux it is a little bit more complicated as you'll need to install php-pear and compile the module with pecl

 

[root@centos ~]# yum install php-pear

[root@centos ~]# yum install php-pecl-memcache


Compile memcache

[root@centos ~]# pecl install memcache

 

2. Test if memcached is properly loaded in PHP


Once installed lets check if memcached service is running and memcached support is loaded as module into PHP core.

 

webserver1:~# ps -efa  | egrep memcached
nobody   14443     1  0 Oct23 ?        00:04:34 /usr/bin/memcached -v -m 64 -p 11211 -u nobody -l 127.0.0.1 -l 192.168.0.1

root@webserver1:/# php -m | egrep memcache
memcached


To get a bit more verbose information on memcache version and few of memcached variable settings:

root@webserver1:/# php -i |grep -i memcache
/etc/php/7.4/cli/conf.d/25-memcached.ini
memcached
memcached support => enabled
libmemcached version => 1.0.18
memcached.compression_factor => 1.3 => 1.3
memcached.compression_threshold => 2000 => 2000
memcached.compression_type => fastlz => fastlz
memcached.default_binary_protocol => Off => Off
memcached.default_connect_timeout => 0 => 0
memcached.default_consistent_hash => Off => Off
memcached.serializer => php => php
memcached.sess_binary_protocol => On => On
memcached.sess_connect_timeout => 0 => 0
memcached.sess_consistent_hash => On => On
memcached.sess_consistent_hash_type => ketama => ketama
memcached.sess_lock_expire => 0 => 0
memcached.sess_lock_max_wait => not set => not set
memcached.sess_lock_retries => 5 => 5
memcached.sess_lock_wait => not set => not set
memcached.sess_lock_wait_max => 150 => 150
memcached.sess_lock_wait_min => 150 => 150
memcached.sess_locking => On => On
memcached.sess_number_of_replicas => 0 => 0
memcached.sess_persistent => Off => Off
memcached.sess_prefix => memc.sess.key. => memc.sess.key.
memcached.sess_randomize_replica_read => Off => Off
memcached.sess_remove_failed_servers => Off => Off
memcached.sess_sasl_password => no value => no value
memcached.sess_sasl_username => no value => no value
memcached.sess_server_failure_limit => 0 => 0
memcached.store_retry_count => 2 => 2
Registered save handlers => files user memcached


Make sure /etc/default/memcached (on Debian is enabled) on CentOS / RHELs this should be /etc/sysconfig/memcached

webserver1:~# cat default/memcached 
# Set this to no to disable memcached.
ENABLE_MEMCACHED=yes

As assured on server1 memcached + php is ready to be used, next login to Linux server 2 and repeat the same steps install memcached and the module and check it is showing as loaded.

Next place under some of your webservers hosted websites under check_memcached.php below PHP code
 

<?php
if (class_exists('Memcache')) {
    $server = 'localhost';
    if (!empty($_REQUEST[‘server’])) {
        $server = $_REQUEST[‘server’];
    }
    $memcache = new Memcache;
    $isMemcacheAvailable = @$memcache->connect($server);

    if ($isMemcacheAvailable) {
        $aData = $memcache->get('data');
        echo '<pre>';
        if ($aData) {
            echo '<h2>Data from Cache:</h2>';
            print_r($aData);
        } else {
            $aData = array(
                'me' => 'you',
                'us' => 'them',
            );
            echo '<h2>Fresh Data:</h2>';
            print_r($aData);
            $memcache->set('data', $aData, 0, 300);
        }
        $aData = $memcache->get('data');
        if ($aData) {
            echo '<h3>Memcache seem to be working fine!</h3>';
        } else {
            echo '<h3>Memcache DOES NOT seem to be working!</h3>';
        }
        echo '</pre>';
    }
}

if (!$isMemcacheAvailable) {
    echo 'Memcache not available';
}

?>


Launch in a browser https://your-dns-round-robined-domain.com/check_memcached.php, the browser output should be as on below screenshot:

check_memcached-php-script-website-screenshot

3. Configure memcached daemons on both nodes

All we need to set up is the listen IPv4 addresses

On Host Webserver1
You should have in /etc/memcached.conf

-l 127.0.0.1
-l 192.168.0.1

webserver1:~# grep -Ei '\-l' /etc/memcached.conf 
-l 127.0.0.1
-l 192.168.0.1


On Host Webserver2

-l 127.0.0.1
-l 192.168.0.200

 

webserver2:~# grep -Ei '\-l' /etc/memcached.conf
-l 127.0.0.1
-l 192.168.0.200

 

4. Configure memcached in php.ini

Edit config /etc/php.ini (on CentOS / RHEL) or on Debians / Ubuntus etc. modify /etc/php/*/apache2/php.ini (where depending on the PHP version you're using your php location could be different lets say /etc/php/5.6/apache2/php.ini):

If you wonder where is the php.ini config in your case you can usually get it from the php cli:

webserver1:~# php -i | grep "php.ini"
Configuration File (php.ini) Path => /etc/php/7.4/cli
Loaded Configuration File => /etc/php/7.4/cli/php.ini

 

! Note: That on on PHP-FPM installations (where FastCGI Process Manager) is handling PHP requests,path would be rather something like:
 

/etc/php5/fpm/php.ini

in php.ini you need to change as minimum below 2 variables
 

session.save_handler =
session.save_path =


By default session.save_path would be set to lets say session.save_path = "

/var/lib/php7/sessions"


To make php use a 2 central configured memcached servers on webserver1 and webserver2 or even more memcached configured machines set it to look as so:

session.save_path="192.168.0.200:11211, 192.168.0.1:11211"


Also modify set

session.save_handler = memcache


Overall changed php.ini configuration on Linux machine 1 ( webserver1 ) and Linux machine 2 ( webserver2 ) should be:

session.save_handler = memcache
session.save_path="192.168.0.200:11211, 192.168.0.1:11211"

 

Below is approximately how it should look on both :

webserver1: ~# grep -Ei 'session.save_handler|session.save_path' /etc/php.ini
;; session.save_handler = files
session.save_handler = memcache
;     session.save_path = "N;/path"
;     session.save_path = "N;MODE;/path"
;session.save_path = "/var/lib/php7/sessions"
session.save_path="192.168.0.200:11211, 192.168.0.1:11211"
;       (see session.save_path above), then garbage collection does *not*
 

 

webserver2: ~# grep -Ei 'session.save_handler|session.save_path' /etc/php.ini
;; session.save_handler = files
session.save_handler = memcache
;     session.save_path = "N;/path"
;     session.save_path = "N;MODE;/path"
;session.save_path = "/var/lib/php7/sessions"
session.save_path="192.168.0.200:11211, 192.168.0.1:11211"
;       (see session.save_path above), then garbage collection does *not*


As you can see I have configured memcached on webserver1 to listen on internal local LAN IP 192.168.0.200 and on Local LAN eth iface 192.168.0.1 on TCP port 11211 (this is the default memcached connections listen port), for security or obscurity reasons you might choose another empty one. Make sure to also set the proper firewalling to that port, the best is to enable connections only between 192.168.0.200 and 192.168.0.1 on each of machine 1 and machine 2.

loadbalancing2-php-sessions-scheme-explained
 

5. Enable Memcached for session redundancy


Next step is to configure memcached to allow failover (e.g. use both memcached on 2 linux hosts) and configure session redundancy.
Configure /etc/php/7.3/mods-available/memcache.ini or /etc/php5/mods-available/memcache.ini or respectively to the right location depending on the PHP installed and used webservers version.
 

webserver1 :~#  vim /etc/php/7.3/mods-available/memcache.ini

; configuration for php memcached module
; priority=20
; settings to write sessions to both servers and have fail over
memcache.hash_strategy=consistent
memcache.allow_failover=1
memcache.session_redundancy=3
extension=memcached.so

 

webserver2 :~# vim /etc/php/7.3/mods-available/memcache.ini

; configuration for php memcached module
; priority=20
; settings to write sessions to both servers and have fail over
memcache.hash_strategy=consistent
memcache.allow_failover=1
memcache.session_redundancy=3
extension=memcached.so

 

memcache.session_redundancy directive must be equal to the number of memcached servers + 1 for the session information to be replicated to all the servers. This is due to a bug in PHP.
I have only 2 memcached configured that's why I set it to 3.
 

6. Restart Apache Webservers

Restart on both machines webserver1 and webserver2 Apache to make php load memcached.so
 

webserver1:~# systemctl restart httpd

webserver2:~# systemctl restart httpd

 

7. Restart memcached on machine 1 and 2

 

webserver1 :~# systemctl restart memcached

webserver2 :~# systemctl restart memcached

 

8. Test php sessions are working as expected with a php script

Copy to both website locations to accessible URL a file test_sessions.php:
 

<?php  
session_start();

if(isset($_SESSION[‘georgi’]))
{
echo "Sessions is ".$_SESSION[‘georgi’]."!\n";
}
else
{
echo "Session ID: ".session_id()."\n";
echo "Session Name: ".session_name()."\n";
echo "Setting 'georgi' to 'cool'\n";
$_SESSION[‘georgi’]='cool';
}
?>

 

Now run the test to see PHP sessions are kept persistently:
 

hipo@jeremiah:~/Desktop $ curl -vL -s http://pc-freak.net/session.php 2>&1 | grep 'Set-Cookie:'
< Set-Cookie: PHPSESSID=micir464cplbdfpo36n3qi9hd3; expires=Tue, 10-Nov-2020 12:14:32 GMT; Max-Age=86400; path=/

hipo@jeremiah:~/Desktop $ curl -L –cookie "PHPSESSID=micir464cplbdfpo36n3qi9hd3" http://83.228.93.76/session.php http://213.91.190.233/session.php
Session is cool!
Session is cool!

 

Copy to the locations that is resolving to both DNS servers some sample php script such as sessions_test.php  with below content:

<?php
    header('Content-Type: text/plain');
    session_start();
    if(!isset($_SESSION[‘visit’]))
    {
        echo "This is the first time you're visiting this server\n";
        $_SESSION[‘visit’] = 0;
    }
    else
            echo "Your number of visits: ".$_SESSION[‘visit’] . "\n";

    $_SESSION[‘visit’]++;

    echo "Server IP: ".$_SERVER[‘SERVER_ADDR’] . "\n";
    echo "Client IP: ".$_SERVER[‘REMOTE_ADDR’] . "\n";
    print_r($_COOKIE);
?>

Test in a Web Opera / Firefox / Chrome browser.

You should get an output in the browser similar to:
 

Your number of visits: 15
Server IP: 83.228.93.76
Client IP: 91.92.15.51
Array
(
    [_ga] => GA1.2.651288003.1538922937
    [__utma] => 238407297.651288003.1538922937.1601730730.1601759984.45
    [__utmz] => 238407297.1571087583.28.4.utmcsr=google|utmccn=(organic)|utmcmd=organic|utmctr=(not provided)
    [shellInABox] => 467306938:1110101010
    [fpestid] => EzkIzv_9OWmR9PxhUM8HEKoV3fbOri1iAiHesU7T4Pso4Mbi7Gtt9L1vlChtkli5GVDKtg
    [__gads] => ID=8a1e445d88889784-22302f2c01b9005b:T=1603219663:RT=1603219663:S=ALNI_MZ6L4IIaIBcwaeCk_KNwmL3df3Z2g
    [PHPSESSID] => mgpk1ivhvfc2d0daq08e0p0ec5
)

If you want to test php sessions are working with text browser or from another external script for automation use something as below PHP code:
 

<?php
// save as "session_test.php" inside your webspace  
ini_set('display_errors', 'On');
error_reporting(6143);

session_start();

$sessionSavePath = ini_get('session.save_path');

echo '<br><div style="background:#def;padding:6px">'
   , 'If a session could be started successfully <b>you should'
   , ' not see any Warning(s)</b>, otherwise check the path/folder'
   , ' mentioned in the warning(s) for proper access rights.<hr>';
echo "WebServer IP:" . $_SERVER[‘SERVER_ADDR’] . "\n<br />";
if (empty($sessionSavePath)) {
    echo 'A "<b>session.save_path</b>" is currently',
         ' <b>not</b> set.<br>Normally "<b>';
    if (isset($_ENV[‘TMP’])) {
        echo  $_ENV[‘TMP’], ‘” ($_ENV[“TMP”]) ';
    } else {
        echo '/tmp</b>" or "<b>C:\tmp</b>" (or whatever',
             ' the OS default "TMP" folder is set to)';
    }    
    echo ' is used in this case.';
} else {
    echo 'The current "session.save_path" is "<b>',
         $sessionSavePath, '</b>".';
}

echo '<br>Session file name: "<b>sess_', session_id()
   , '</b>".</div><br>';
?>

You can download the test_php_sessions.php script here.

To test with lynx:

hipo@jeremiah:~/Desktop $ lynx -source 'https://pc-freak.net/test_php_sessions.php'
<br><div style="background:#def;padding:6px">If a session could be started successfully <b>you should not see any Warning(s)</b>, otherwise check the path/folder mentioned in the warning(s) for proper access rights.<hr>WebServer IP:83.228.93.76
<br />The current "session.save_path" is "<b>tcp://192.168.0.200:11211, tcp://192.168.0.1:11211</b>".<br>Session file name: "<b>sess_5h18f809b88isf8vileudgrl40</b>".</div><br>

Briefly unavailable for scheduled maintenance. Fix WordPress after interrupted upgrade

Thursday, March 2nd, 2017

Reading Time: 4minutes

briefly-unavaiable-for-a-scheduled-maintenance-wordpress-website-fix-howto_1

I've recenty tried Update my WordPress blog sites and being unattentive I've selected all the plugins possitble for Upgrade by checking the "Select All" check box on the Update dialogs and almost automatically int he hurry pressing Update button however out of a sudden I've realized I could screw up my websites brutally as some of the plugins to upgraded might be lacking 100% compitability with their prior versions.

I've made a messes out of my blog many times during upgrades because of choosing to upgrade the wrong not 100% compatible plugins and I know well how painful and hard to track it could be a misbehaving incompatible plugin or how ot could cause a severe sluggishness to blog which automatically reflects on how well the website search engine ranked in Google / Yahoo / Bing indexed etc.

Thus as an almost unconcsious reaction to prevent myself the future troubles I've tried to cancel the update request in Firefox browser and trying to reload the Update page with a hope that I might be quick enough for the Apache / WP / MySQLbackend Update Update queries request to be delaying for processing but I was too slow and bang! I ended up with the following unpleasent message in my browser:

briefly-unavaiable-for-a-scheduled-maintenance-wordpress-website-fix-howto
Briefly unavailable for scheduled maintenance.
 

As you could guess that message caused me quite a lot of worries at hand especially since I've already break up my sites many times by doing quick unmindful reactions and the fact that there is Google Adsense ads appearing which does give me some Return on Investment cents every now and then …

It took me few minutes of research online to find what really happened and how to fix / resolve the WebSites normal operations.
 


So what causes the Briefly unavailable for scheduled maintenance. appears ?

When WordPress does some of its integrated maintenance jobs a plugin enable / disable or any task that has to modify crucial configurations inside the database WordPress does disable access to all end clients to itself in order to protect its sensitive data to appear to browser requestors as showing some unexpected information to end client browser could be later used by crackers / hackers or a possibly open a security hole for an attacker.

The message is wordpress generated notice and it is pretty normal for the end user to see it during the WP site installation update depending on how many plugins are installed and loaded to the site and how long it will take for the backend Linux / Windows server to fetch the archived .zips of plugins and substitute with the new ones and update the files extracting them to wp-content/plugins and updating the respective required SQL database / tables it could be showing for end users from few secs to few minutes.

However under some circumstances on Browser request timeout to remote wordpress site due to a network connectivity issue or just a bad configuration of Apache for requests timeout (or a slow remote server Apache responce time due to server Hardware / Mem overload) or a stupid browser "Stop" / cancel request like in my case you end up with the Briefly unavailable for scheduled maintenance and you can can longer access the your https://siteurl.com/wp-admin Admin Panel.

The message is triggered by a WP craeted file .maintenance inside /var/www/blog-site/ e.g. WordPress PHP scripts does check for/var/www/blog-site/.maintenance
existence and if it is matched the WP scripts does generate the Briefly unavailble … message.


How to resolve the "Briefly unavailable for scheduled maintenance. Check back in a minute" WordPress error ?

As you might guess removing the maintenance "coming soon" like message in most of the cases comes to just deleting the .maintenance file, to do so:

1. Login to remote server via FTP or SFTP
2. Locate your WP website root folder that should be something like /var/www/blog-site/.maintenance and issue:
issue something like:
 

$ rm -rf /var/www/bog-site/.maintenance


Assuming that some plugin Update .zip extraction or SQL update query did not ended being half installed/ executedthat should solve the error.
To check whether all is back to normal just refresh your browser pointing to the "broken" site. If it appears well you can thank God for that 🙂
If not check the apache error logs and php error logs and see which of the php scripts is failing and then try to manually fetch and unzip the WP .zip package to wp-content/plugins folder and give it another try and if God bless so it will work as before 🙂

How to prevent your WP based business in future from such nasty errors using A Staging site (test) version of your blog ?

Just run a duplicate of your website under a separate folder on your hosting and do enable the same plugins as on the primary website and copy over the MySQL / PostgreSQL Database from your Live site to the Staging, then once it is enabled before doing any crucial WordPress version updates or Plugins Update always do try the Upgrade first on your Test Staging site. If it does execute fine there in most of the cases the result should be the same on the Production host and that could solve you effors and nerves of debugging a hard to get failure errors or faulty plugins without affecting what your End users see.

If you're not hosting the WordPress install under your own hosting like me you can always use some of the public available hostings like  BlueHost or WPEngine

 

Check and Restart Apache if it is malfunctioning (not returning HTML content) shell script

Monday, March 19th, 2012

Reading Time: 2minutes

Check and Restart Apache Webserver on Malfunction, Apache feather logo

One of the company Debian Lenny 5.0 Webservers, where I'm working as sys admin sometimes stops to properly server HTTP requests.
Whenever this oddity happens, the Apache server seems to be running okay but it is not failing to return requested content

I can see the webserver listens on port 80 and establishing connections to remote hosts – the apache processes show normally as I can see in netstat …:

apache:~# netstat -enp 80
Active Internet connections (w/o servers)
Proto Recv-Q Send-Q Local Address Foreign Address State User Inode PID/Program name
tcp 0 0 xxx.xxx.xxx.xx:80 46.253.9.36:5665 SYN_RECV 0 0 -
tcp 0 0 xxx.xxx.xxx.xx:80 78.157.26.24:5933 SYN_RECV 0 0 -
...

Also the apache forked child processes show normally in process list:

apache:~# ps axuwwf|grep -i apache
root 46748 0.0 0.0 112300 872 pts/1 S+ 18:07 0:00 \_ grep -i apache
root 42530 0.0 0.1 217392 6636 ? Ss Mar14 0:39 /usr/sbin/apache2 -k start
www-data 42535 0.0 0.0 147876 1488 ? S Mar14 0:01 \_ /usr/sbin/apache2 -k start
root 28747 0.0 0.1 218180 4792 ? Sl Mar14 0:00 \_ /usr/sbin/apache2 -k start
www-data 31787 0.0 0.1 219156 5832 ? S Mar14 0:00 | \_ /usr/sbin/apache2 -k start

In spite of that, in any client browser to any of the Apache (Virtual hosts) websites, there is no HTML content returned…
This weird problem continues until the Apache webserver is retarted.
Once webserver is restarted everything is back to normal.
I use Apache Check Apache shell script set on few remote hosts to regularly check with nmap if port 80 (www) of my server is open and responding, anyways this script just checks if the open and reachable and thus using it was unable to detect Apache wasn't able to return back HTML content.
To work around the malfunctions I wrote tiny script – retart_apache_if_empty_content_is_returned.sh

The scripts idea is very simple;
A request is made a remote defined host with lynx text browser, then the output of lines is counted, if the output returned by lynx -dump http://someurl.com is less than the number returned whether normally invoked, then the script triggers an apache init script restart.

I've set the script to periodically run in a cron job, every 5 minutes each hour.
# check if apache returns empty content with lynx and if yes restart and log it
*/5 * * * * /usr/sbin/restart_apache_if_empty_content.sh >/dev/null 2>&1

This is not perfect as sometimes still, there will be few minutes downtime, but at least the downside will not be few hours until I am informed ssh to the server and restart Apache manually

A quick way to download and set from cron execution my script every 5 minutes use:

apache:~# cd /usr/sbin
apache:/usr/sbin# wget -q https://pc-freak.net/bscscr/restart_apache_if_empty_content.sh
apache:/usr/sbin# chmod +x restart_apache_if_empty_content.sh
apache:/usr/sbin# crontab -l > /tmp/file; echo '*/5 * * * *' /usr/sbin/restart_apache_if_empty_content.sh 2>&1 >/dev/null