The Wonderful World of Linux A mostly dead cpanel/linux blog

Change PECL tmp Directory

If you’ve ever ran into the “/tmp/foobar does not exist or is not executable” after running “pecl install foobar” then its likely because /tmp is set to noexec in the fstab. Obviously removing the noexec flag and rebooting or remounting will fix the problem but generally the time required to do this and the fact that processes will need to be temporarily halted doesnt make this necessarily the best option. Lets do a quick little fix!

mkdir /root/tmp
pecl config-set temp_dir /root/tmp

If this yells at you and says it failed, try this:

pear config-set temp_dir /root/tmp

Then “pecl install foobar”!

Find Which Accounts are Potential Spammers in cPanel/Exim

So you’ve discovered that all of a sudden your server load has shot and your email inbox is getting filled up with hundreds of bounce backs. You sir may be spamming! Now if you are a spammer, this isnt really much of a shock.

But if you’re not a spammer you may be wondering what happened. Well odds are your website got hacked or your personal machine has a virus/malware. Now usually if your website gets hacked its because you are using an outdated version of your CMS software. Because wordpress doesn’t auto update and we don’t always login every day (especially if your wordpress just hosts a static website) it can be hard to keep up with the constant updates. And of course in doing so all the little hackers out there are now able to exploit whatever security holes you didn’t patch. On top of this it isn’t just the wordpress core we have to worry about, but also the plugins and the themes. Joomla and drupal and practically all other CMS’s follow the same logic. Keep your apps up to date and your chances of being compromised slim down quickly. What happens though when you are compromised? Usually the attacker places a php file on the server that acts as part of a ddos or a script that sends out a ton of spam. If your personal machine was compromised then whatever application you are running to connect to your email (such as outlook, thunderbird, etc) is usually used (or they just grab any IMAP/SMTP connection info) and use that to start spamming.

Ok, so we know that someone on the server is spamming. We don’t know if its a script or if its because someone’s personal machine got attacked. Lets take a look a couple one liners to help out with this. First, lets look at a command which searches for all external logins (meaning the personal local machine was compromised)

$ exigrep @ /var/log/exim_mainlog | grep _login | sed -n 's/.*_login:\(.*\)S=.*/\1/p' | sort | uniq -c | sort -nr -k1
1 [email protected]
3 [email protected]
59 [email protected]

So this will exigrep through our mail log and return any line containing an @ (meaning pretty much everything) cut out the dovecot_login or courier_login (whichever one you use) and then sort it and count how many instances there are. In this case you can see that the email account [email protected] is sending much, much more than the other two email accounts it found. This doesn’t immediately mean that its a spamming account, it could be legitimate of course but it gets you on the right path.

Now lets look at a few one liners for checking which user/account has been hacked:

$ exigrep @ /var/log/exim_mainlog | grep U= | sed -n 's/.*U=\(.*\)S=.*/\1/p' | sort | uniq -c | sort -nr -k1
3 user1 P=local
74 user2 P=local

So here the user “user2” is sending the most email on the system so we know that this user is likely responsible for the spam. Lets see if we can track down the script!

grep "cwd=" /var/log/exim_mainlog | awk '{for(i=1;i<=10;i++){print $i}}' | sort |uniq -c| grep cwd | sort -n | grep /home/

Running this will look at any lines in the exam log that contains the “cwd” string. This should help narrow it down the folder where the spam is happening. But we can get even more specific! Note that this command doesn’t have as high of a success rate as the previous ones but when it works it saves so much headache.

grep X-PHP-Script /var/spool/exim/input/*/*-H | awk '{print $3}' | sort | uniq -c | sort -nr

Now unlike the others this actually searches the active email queue. So if you have hundreds or thousands of email queued up (and you can check this by running exam -bpc) this should work. It looks for the X-PHP-Script field in the header of the emails. This should be enabled by default in cpanel, if not it can be enabled in the whm. But anyways this should again sort and count exactly which script sent the email! Pretty cool, right!

So there you have it, if you are unfortunate enough to have a compromised system this will better help identify where the problem lies. And once you know, you can help fix it and safe guard yourself for the future. In addition to the wordpress and CMS tips above, you may want to look at even more security oriented plugins:


Ive seen this used quite a bit over the interwebs and although Ive always been on the problem side of it, it should give you a little extra security to prevent such attacks. For PC based compromises, you gotta have a virus scanner and a malware scanner, here are my favorites:

Windows Defender MalwareBytes

People will argue that you use avast or avg over windows defender, personally I don’t think its necessary. Windows Defender generally does a good job at actively scanning your computer and catching anything that comes through. Its quiet, sits in the background, integrates great with windows and is completely free (and even free for businesses up to 10 computers)! The second is malwarebytes and Ive seen it personally catch more malware than any competitor. Its completely free but does have a paid option for more and worthwhile features.

“Is there anything else I can do?” Uh yeah! Stop them before they even get to your server! Thats where the following tools come into play:

6Scan CloudFlare

6scan is a paid product and will scan and even fix many of the vulnerabilities you encounter. Rather than finding out your hacked by a visitor, a bounce back or the google malware page, 6scan will not only alert you but also help fix the issue! Cloudflare, in addition to speeding up your website through their CDN, it also includes a firewall to help block many of the known attackers out there. Oh, and its free with paid versions available for higher speed, better analytics and more finely tuned security settings.

Hopefully after you read this you know how to not only identify hacked accounts and spamming accounts but also know a few more steps to help prevent the same thing from happening in the future. I would love to hear about some of your stories or strategies so please leave them in the comments below!

Installing Git on cPanel/CentOS Servers

Git is awesome, plain and simple! If you would like to install this on a cPanel and/or a CentOS based server you have 2 options: Installing from and RPM or installing from source.

RPM/Yum Based

You should just simply be able to do a yum install on the server and git will e pulled from the “updates” repo:

yum install git

If you run into a dependency issue with perl-Git, check and make sure that perl is not entered into the excludes in /etc/yum.conf. Now one issue with simply pulling from the basic repo is that the version may not be as up to date as possible. On my server I pulled down version 1.7.1, however the latest stable version of git is 1.8.2. You may be able to find much more up to date versions in the EPEL repo or RPMForge. You can also install the latest available version straight from source:

Source Install

First off, you’ll want to make sure you have any and all dependencies taken care of:

yum -y install curl-devel expat-devel gettext-devel openssl-devel zlib-devel

Now lets grab the latest version and git (See what I did there?) to work!!

cd /usr/local/src/
tar xzf master
cd git*
make prefix=/usr/local all
make prefix=/usr/local install 

And thats it! Now feel free to learn git ( and give it a shot!

[email protected] [~]# git --version
git version 1.8.2.GIT

Rebuilding Spamd on cPanel

Spamd on cpanel seems to always fail for me. I dont if its something special within my config or if its just a cpanel thing in general. Because I am a lazy admin I dont always bother with fixing things but jump to rebuilding them…this usually does the trick for me:

rpm -e `rpm -qa | grep spam`
/usr/local/cpanel/scripts/installspam --force
sa-learn -D --force-expire
sa-update -D

This just completely reinstalls spamd and cleans up any messes. Does the trick for me!

Getting Started with HA Proxy [High Availability]

HAProxy is a very fast, very awesome and very free TCP/HTTP Load Balancer and Failover program. Its very simplistic but also quite powerful. The configuration is flexible enough to fit into many high traffic infrastructures but also simple enough to fit into any design seeking simple high availability. It can support tens of thousands of connections at high speeds and can even private a private to public address “translation” if you will by hiding web servers from the internet.

One of the most popular scenarios for HA Proxy is load balancing in its simplest form. A web server struggling to keep up with traffic demands would benefit by throwing the software in front of the web server, adding a second server and balancing between the two. This would effectively cut the load in half on any individual server and distribute it amongst the two, or however many you want, and should yield a much more stable web site.

However load balancing is only the beginning, because of the powerful nature of the config many webmasters have found that much of its value lies in its ability to be a “traffic cop” if you will. Imgur has a really interesting post on how they utilize the software, not for load balancing, but for directing any image requests to another server. Essentially, HA Proxy is grabbing any GET requests destined for any JPG, PNG or whatever and sending to a separate Lighttpd process and everything else goes to apache. It would be trivial to take this concept and use it to your advantage. Rather than writing any custom code to separate out certain elements you could simply write your requests in HA Proxy to rewrite your images, javascripts, css files, whatever to a CDN and send all traffic to a dedicated server. Its the flexibility and power that makes this a necessary piece of software for any linux fan to take hold of. There are dozens of other case uses (Conveniently located on their site) and quite a few more advanced features. For now, lets just get this software installed and setup a basic round robin config setup!

The Setup

The setup that we will be working with is populated with 1 HAProxy box and 2 web servers all on a single LAN. You can see the layout in this image:

HAProxy Example Layout

All configuration will be done on the HAProxy machine and it is assumed that the web servers are already configured and displaying correctly. The HAProxy box I am using is running CentOS 5. Lets get started!


If you are using CentOS, you will need to enable the EPEL repo and then simply install it via yum:

yum install haproxy

Debian should have this package in the default repo:

 sudo apt-get install haproxy

If you are like me and love installing things from source, first make sure you have gcc installed. If you’re on CentOS, I just like installing all of the development tools. If you are low on space or resources you may not want to install everything but its just easier in my opinion:

yum groupinstall "Development Tools"

The download and compile the program:

cd /usr/local/src/
tar xzf haproxy-1.4.22.tar.gz
cd haproxy-1.4.22
make TARGET=linux26 #Change TARGET accordingly
cp haproxy /usr/sbin/

On the fifth line, make sure that you change the TARGET to your linux version. If you are unsure, run the following and record the first 2 numbers:

uname -r

You can also append “ARCH=i386” to build the 32 bit binary if you are running on a 64 but system. HAProxy works on quite a few machines, you have the following options when it comes to the TARGET variable:

  • linux22 for Linux 2.2
  • linux24 for Linux 2.4 and above (default)
  • linux24e for Linux 2.4 with support for a working epoll (> 0.21)
  • linux26 for Linux 2.6 and above
  • linux2628 for Linux 2.6.28 and above (enables splice and tproxy)
  • solaris for Solaris 8 or 10 (others untested)
  • freebsd for FreeBSD 5 to 8.0 (others untested)
  • openbsd for OpenBSD 3.1 to 4.6 (others untested)
  • aix52 for AIX 5.2
  • cygwin for Cygwin
  • generic for any other OS.
  • custom to manually adjust every setting

You can launch the haproxy by running the following command:

haproxy -f /etc/haproxy/haproxy.cfg -sf `pgrep haproxy`

This can be ran multiple times without fear of spawning overlapping processes. The “-sf pgrep haproxy” will kill off the previous command if it exists and launch HAProxy using /etc/haproxy/haproxy.cfg as its config, which we will configure in the next section.

You can also create an init script from the file included:

cp examples/haproxy.init /etc/init.d/haproxy
chmod a+x /etc/init.d/haproxy

Note that this will not run until /etc/haproxy/haproxy.cfg is configured…so lets do that!

Build a Basic Config

We are going to assume that your config is held in /etc/haproxy/haproxy.cfg. Please note that if you installed this with a precompiled binary such as through a deb package or RPM this may be different. For us though, we need to first create the /etc/haproxy folder, once created touch a file called haproxy.cfg. Open the file with your favorite editor and paste the following:

        log     global
        mode    http
        option  httplog
        option  dontlognull
        retries 3
        option redispatch
        maxconn 20000
        contimeout      5000
        clitimeout      50000
        srvtimeout      50000

listen webservers
       mode http
       stats enable     #Optional: Enables stats interface
       stats uri /haproxy?stats     #Optional: Statistics URL
       stats realm Haproxy\ Statistics
       stats auth haproxy:stats     #Optional: username:password of site
       balance roundrobin     #balance type
       cookie webserver insert indirect nocache 
       option httpclose
       option forwardfor     #Add X-ForwardFor in the header
       server server1 cookie server1 weight 1 check
       server server2 cookie server2 weight 1 check

Modify this if needed and save it to /etc/haproxy/haproxy.cfg. Once you’re finished we can actually check the syntax of this config file by running the following:

haproxy -f /etc/haproxy/haproxy.cfg -c

If its good, you should see “Configuration file is valid”. Go ahead and start everything up!

/etc/init.d/haproxy start

Go ahead and visit in your web browser and you should see the page from one of your two web servers! Pretty cool right?!

Config Commentary

I added a few comments to some of the lines within the config to help better explain what it does. Here is a little more info on some of those commands.

  • Stats Commands
    • stats enable - Simply enables the statistics web interface
    • stats uri /haproxy?stats - This defines the url where you can visit the stats. ex:
    • stats realm Haproxy\ Statistics - The site info in the authentication box
    • stats auth haproxy:stats - The username and password of the url above. If this is removed the stats site will not require authentication
  • balance roundrobin - The method used to balance between the two servers. Here is a list of the more popular options below, the full list can be found in the documentation
    • round robin - Each server is used in turns, according to their weights. This is the smoothest and fairest algorithm when the server’s processing time remains equally distributed.
    • source - The source IP address is hashed and divided by the total weight of the running servers to designate which server will receive the request. This ensures that the same client IP address will always reach the same server as long as no server goes down or up. If the hash result changes due to the number of running servers changing, many clients will be directed to a different server.
    • leastconn - The server with the lowest number of connections receives the connection. Round-robin is performed within groups of servers of the same load to ensure that all servers will be used. Use of this algorithm is recommended where very long sessions are expected, such as LDAP, SQL, TSE, etc… but is not very well suited for protocols using short sessions such as HTTP
  • server server1 cookie server1 weight 1 check - There are a lot of options you can use here. Lets break some down
    • cookie [value] - Sets the name of the cooke to store on the client. In this case its named server1
    • weight [int] - This, to put it simply, puts a specific value on the servers power/importance. In our scenario both are valued at 1 meaning that both are taken as equally powerful and thus should be equally distributed. The default is 1, the max is 256. The higher the number the more load on the server proportional to the values of the other servers.
    • backup - If this value is added in the server line, it is disabled unless all non backup servers go down. This is good if you want to only use this as a failover.
    • check - This enables health checks on the server. If this is not added a server will always be considered available even if it actually is not physically up.
    • disabled - When added the server is marked as down in maintenance mode
    • fall [int] - States how long to wait until the server is considered to be down. Default is 3
    • rise [int] - States how long to wait until the server is backup. Default is 2

There are many more options and more explanations which can be found in the documentation.

What Next?

About that documentation, you can find it here. There are far too many options to detail here, this being a beginners guide. There are some more examples and commands that are pretty cool and should help you out!

  • monitor-uri /haproxy - When this is enabled, visiting it simply provides a success 200 if the haproxy service is working
  • option httpchk /check.html - Essentially its a much more powerful version of the “check” command. It checks the backend servers and checks /check.html. It its anything but a valid 200 response, it marks the server as down. You can take this a step further and create a php page that check for apache, mysql, server load, etc… and if any one of those is invalid it sends an error 500. This will take the server down until you can correct it.

ACL’s and Frontends

I wanted to wrap this up with another sample configuration. This one actually incorporates one of my favorite features and that’s front ends and ACL’s. Take a look ad this:

	#maxconn 30000
	#ulimit-n 655360
	#chroot /home/haproxy
	#uid 500
	#gid 500
	#nbproc 1
	pidfile /var/run/
	log     local4 notice

    mode http

    clitimeout          600000   # maximum inactivity time on the client side
    srvtimeout          600000   # maximum inactivity time on the server side
    timeout connect     8000  	 # maximum time to wait for a connection attempt to a server to succeed

    stats enable
    stats auth		admin:password
    stats uri		/admin?stats
    stats refresh	5s
    stats hide-version
    stats realm "Loadbalancer stats"
    #option httpchk	GET /status
    retries		5
    option redispatch
    option forwardfor
    option httpclose

    monitor-uri /test

    balance roundrobin	# each server is used in turns, according to assigned weight

    default_backend default

frontend http
    bind :80

    acl nagios path_reg ^/nagios/?
    acl app path_reg ^/app/?
    acl mysite path_reg ^/blog/?

    use_backend nagios if nagios
    use_backend app if app
    use_backend mysite if mysite

backend default
    server web1 cookie A check inter 1000 rise 2 fall 5 maxqueue 50 weight 30
    server web2 cookie B check inter 1000 rise 2 fall 5 maxqueue 50 weight 75
    stats admin if TRUE
    option httpchk HEAD /check.txt HTTP/1.0
    errorfile 503 /var/www/html/503.html

backend nagios
    server nagios check inter 1000 rise 2 fall 5 maxqueue 50 weight 30
    errorfile 503 /var/www/html/503.html

backend app
    server clear check inter 1000 rise 2 fall 5 maxqueue 50 weight 30

backend mysite
    server mysite check inter 1000 rise 2 fall 5 maxqueue 50 weight 30

The defaults section are a set of instructions that apply to all backend servers. Really useful if you have a few set of instructions that need to apply to everything. In the ACL section for “nagios” we see a regex value that matches /nagios. Then below it a command says that if the incoming request matches the regex value, then pass it to the “nagios” backend! So if someone hits, HAProxy will display data from! You can also match GET requests as stated in the beginning and match images, javascript files or whatever and use an ACL to route them wherever you want!

HAProxy’s config is really easy to use and it has a lot of value in a web stack. I hope this has given you at least the basics of HAProxy and that you have found some value in it. Enjoy!