When you need to establish connection with a hostname, that has a dynamic IP address (most typical situation – a hostname, that has an unstable Internet connection), Dynamic DNS becomes a simple yet stable solution.

When a hostname obtains an IP address, it is automatically sent to Dynamic DNS system, but users are able to connect using a symbolic name, that remains unchanged (for example, hostname.com).  In order to prevent IP caching, Dynamic DNS records have low TTL values (e.g. 60 seconds).

We’re going to cover the process of using such a configuration with a pppd connection, this package is widely used in Linux with almost all types of Internet connections.

We will take dyndns.org as a Dynamic DNS provider (you’re welcome to select another one: there is no real difference between them). We won’t describe the process of signing up and creating DNS records, but you’re always welcome to follow this guide, and ask us if anything is unclear, though it should be.

We will use inadyn for our connections. It gives the possibility to have your own fixed hostname registered on the internet, although your IP might be changing. It checks periodically whether the IP address stored by the DNS server is the real current IP address of the machine that is running INADYN. Though there are some limitations, that are described here, it is a simple solution, that doesn’t require any specific knowledge. It is a free software, that allows to connect with many Dynamic DNS providers, and you can obtain it by following this link.

You will need to download it, unpack and compile using make.

We will use it in the following way:

inadyn is started when the connection is being established (pppd acquires an IP address and creates an interface). Immediately after that it tries to perform an update and then remains in memory to periodically check for updates. If inadyn is already present in memory, it will be terminated before running a new instance.

In order to do that, we need to touch /etc/ppp/ip-up.local file (create a new one if it doesn’t exist) with the following contents:

#!/bin/sh
kill `ps -A |grep inadyn |sed ’s/[ |\t].*$//’`
sleep 1
/etc/ppp/inadyn -u login -p xxxxx -a lampdocs.dyndns.org –background –update_period_sec 3600

Replace xxxxx with your password, login with your username, obtained at dyndns.org and lampdocs.dyndns.org with a symbolic name you’ve set for your host with dynamic IP.

You can change kill `ps -A |grep inadyn |sed ’s/[ |\t].*$//’` if you don’t like it. Here is another variant:

#!/bin/sh
pkill inadyn
sleep 1
/etc/ppp/inadyn -u login -p xxxxx -a lampdocs.dyndns.org –background –update_period_sec 3600

That’s enough for configuration. Inadyn will be automatically started when a new pppd connection is established. You might need to enable inadyn requests in firewall, so it could connect using http without any problems. Inadyn requires just 2 addresses to connect, and both of them have fixed IPs:

checkip.dyndns.org (IP 208.78.70.70)
members.dyndns.org (IP 204.13.248.112)

You can enable these requests by issuing the following commands:

iptables -t filter -I OUTPUT -p tcp -d 208.78.70.70 –destination-port 80 -j ACCEPT
iptables -t filter -I OUTPUT -p tcp -d 204.13.248.112 –destination -port 80 -j ACCEPT

Then you should input

iptables -t filter -I INPUT -p tcp -s 208.78.70.70 –source-port 80 -j ACCEPT
iptables -t filter -I INPUT -p tcp -s 204.13.248.112 –source-port 80 -j ACCEPT

Or

iptables -t filter -I INPUT -m state –state ESTABLISHED,RELATED -j ACCEPT

- they have the same functionality for our needs. That’s all, Dynamic DNS should be working for you. Note, that you will have to configure inadyn if you choose another Dynamic DNS provider, the configuration is not complicated and you can find it in inadyn documentation.

Please, let us know if you have any problems using Dynamic DNS.

First of all, we will need to download and install GeoIP. We will take free version, but if you really require good stats, you may buy one. Let’s switch to a temporary directory to issue the following commands:

wget http://maxmind.com/download/geoip/api/c/GeoIP.tar.gz;
tar xzvf GeoIP.tar.gz;
cd GeoIP-1.4.6;
./configure;
make;
make install;

You might need to replace 1.4.6 with your version of Geoip, as 1.4.6 is the latest one for today.

Then we need to install Geoip using cpan. I suppose, that you are already familiar with cpan, that’s why I won’t describe initial span setup procedure here. In order to install it, we need to enter cpan from command line. Then issue the following command:

install Geo::IP

Then type quit. And the final stage: we will add Geoip support to awstats.conf. You’re welcome to do it with any editor you like, here is a command line variant using sed.

sed -i ’s/#LoadPlugin=”geoip GEOIP_STANDARD \/pathto\/GeoIP.dat”/LoadPlugin=”geoip GEOIP_STANDARD \/usr\/local\/share\/GeoIP\/GeoIP.dat”/’ /etc/awstats/*.conf

You’re done now. Have fun with your updated awstats!

How do those files usually look like? For example, there was archive.rar before. When it is splitted into 3 parts, we will have archive.rar.000, archive.rar.001, archive.rar.002. How do we join them together?

There is a simple command line solution for this:

cat archive.rar.* > archive.rar

This will join all the parts to a single rar file. Then you fill be able to unrar it. No matter, what was the file type, this is the easiest way to join files together, if they were splitted before.

Let’s determine when we need to have this file on our machine.When we’re compiling any “old” package from sources, we may receive the following error:

error: stropts.h: No such file or directory

What could be the reason and why isn’t this file included into our Linux distribution?This error means that your system doesn’t support STREAMS.

Linux doesn’t support STREAMS (many years ago it was available as a third party module, but it hasn’t worked for years). stropts.h is part of a POSIX XSR option, which is not supported in modern Linux distributions. Do we really need it in Linux?

This means that software you’re trying to compile, will not use the functions listed under stropts.h as they’re not supported by the operating system. So we will do a simple trick that will let you to compile your software without these functions.

Since the stropts.h is required for a successful compilation, the most simple way to solve the issue is to create a blank file named stropts.h under /usr/include. You may want to put any comments there, this way you won’t forget what was the reason to create this file.

This simple trick will help you to compile pptp (1.7.2),  pppd (2.4.5), and I think that the list will be much bigger. That was just my experience, but the sense remains the same: you need this file for a successful compilation. Just create it and have fun!

1. Ensure you’ve got ntp installed. Run rpm -qa | grep ntp to find out if it already exists on your system and yum install ntp if it is not present.

2. Run ntpdate. On CentOS, you will probably get this error:

[root@server ~]# ntpdate
29 Jan 09:52:54 ntpdate[8463]: no servers can be used, exiting

This means you have to include a server into your request, just like shown below.

[root@server ~]# ntpdate 0.pool.ntp.org
29 Jan 10:08:07 ntpdate[8853]: step time server 213.198.55.2 offset 676.605025 sec

This means my server had a 10-minute time diffrence with the time server. After this command my time has been adjusted.

3. Add time synchronization to your crontab and run it as often as you need.

30 * * * * /usr/sbin/ntpdate -s 0.pool.ntp.org

Will sync your clock every hour; I think is enough to perform synchronization weekly.  Let’s do it.

30 10 5 * * /usr/sbin/ntpdate -s 0.pool.ntp.org

This simple guide will keep your clock synchronized. Have fun!

/etc/mail/local-host-names is a text file, that contains the list of domain names and it is taken by sendmail when checking outgoing mail. You should make sure you have this file in the following format:

# local-host-names sample file
lampdocs.com
test.lampdocs.com
# end of file

Please, note, that you should leave the comment on the last line as it is ignored by sendmail. Your local-host-names should have the last line either blank, either commented. Hope this saves someone some time, as this feature is not so well documented and you might have problems with the last domain in the list.

if you receive command not found error, it usually means you have no slocate package installed. On most modern linuxes you can install it using your package manager (yum, apt-get, etc). You are also welcome to compile it from source, that can be found at http://slocate.trakker.ca/ . I don’t think you should have problems with this package; you will then be able to run updatedb and locate.

It is not recommended to run updatedb if you have lots of files on your machine. In this case updatedb execution can slow down your server. But if the number of files is not so high, locate is probably the fastest way to find the files you need.

cat `ls /etc/*{-,_}{release,version} 2>/dev/null | head -n 1`

This will show you something like

The picture below shows the difference between common commands: one shown above and uname -a. Sometimes you don’t need the kernel version, just the operating system name. This command should help you to do that.

When you’re a system administrator, you should often deal with logs. On a Linux server logs are permanently updated, that’s why we need to see last written lines of such a file. If you open it useing cat or more, you won’t see how it changes. A simple command will help you to do this.

tail -f error_log

This will show you last lines of error_log file. For a complete list of options you’re welcome to go here: or just type man tail in your console.

Hope this command will help you to monitor your server and increase its uptime

yum install gcc-c++

You might also want to install it with some more useful packages. Then your command will be:

yum install gcc-c++ compat-gcc-32 compat-gcc-32-c++

Hope this helps you to install g++ under CentOS