droptips.com

There’s a command for that!

So everyone in the Linux/BSD world has used the command “cat” to print out the contents of a file to screen. Infact, I find it hard to believe you can use Linux or BSD properly without using that command on log files and any other text-based file.

But say you have a 1MB text file – an apache logfile for example – that you want to print the contents of to the screen, but you’d rather see the newest contents first – how do you do that?  You certainly don’t want to cat the file and wait until the end…

Well, it’s simple.. what’s the opposite of cat?  tac of course!

From the tac manpage:  tac – concatenate and print files in reverse

So, the next time you want to see the contents of a text file, but want to see the newest content first, tac is the command you need:

tac filename.log

Or, as you’ll probably need (to stop the output just flooding to the screen), you can pipe it to more, just as you can a normal cat.

tac filename.log | more

Whilst quite cool, it can also be quite important to keep a log of the bandwidth being used by a server – whether it’s because you have bandwidth limitations from a data center or ISP, whether you just want to see how much the server uses, or even to detect when something “strange” is going on, basic stats are good – you may not need a fully fledged MRTG setup with graphs though, and this is where vnstat comes in – it’s a basic text-based application, which monitors traffic in and out of a network card.

I install vnstat as a matter of course on all servers I installed – it’s light, and it’s good just to keep a log on things.

For this example, I’ve installed vnstat on an Ubuntu Server, so I’ll be using apt-get to install the package from the standard repositories:

apt-get install vnstat

Once installed, we need to initialise the database.. and in my case, it is for eth0 (you can change eth0 to whichever interface you want to monitor):

vnstat -u -i eth0

Now vnstat will start collecting stats, and you have a whole range of command line options to show different types of stats:

 -d,  --days           show days
 -m,  --months         show months
 -w,  --weeks          show weeks
 -t,  --top10          show top10
So, for example, if we do -m to show monthly stats:
 eth0  /  monthly
   month         rx      |      tx      |   total
-------------------------+--------------+--------------------------------------
  May '09       3.81 MB  |     1.02 MB  |     4.83 MB
  Jun '09     195.68 MB  |   198.24 MB  |   393.92 MB   %%%%%%%%%%%:::::::::::
  Jul '09     217.62 MB  |   103.79 MB  |   321.41 MB   %%%%%%%%%%%%:::::
-------------------------+--------------+--------------------------------------
 estimated       972 MB  |      461 MB  |     1.40 GB
You’ll notice a nice feature at the bottom of estimated use for the next month – again, this takes current/past stats, and works out roughly what it thinks will be used – so in this case, it’s predicting a big increase for July.. which looks about right, especially as it’s used 321MB already.
So, give vnstat a try – I think you’ll like it.
You can read more about vnstat on the official web site:  http://humdi.net/vnstat/