My Linux Brags

Linux File System Hierarchy

2009-07-19T15:07:00.001-07:00

Because everything in Linux is a file, you, as an administrator, are ultimately going to have to deal with files. This makes it very important to understand the Linux file system hierarchy. Most Unixes/Linuxes adhere to the Filesystem Hierarchy Standard which defines a universal/standard directory structure for a Linux/Unix operating system.

Here is a brief overview of the most commonly available directories on a standard Linux system and their purpose:

/bin = houses basic binaries like ls, cat, mv, date, kill, and ps etc.
/boot = houses boot and kernel-specific files.
/dev = houses device files.
/etc = houses host-specific configuration files.
/home = houses user-specific home directories.
/root = home directory of root user.
/lib = houses shared library files used by various programs.
/lost+found = contains files/directories that have gone corrupted, items referred to by bad inodes, lost file fragments and the likes.
/mnt = houses mount points for various removable devices like CD and Floppy.
/opt = houses add-on applications (StarOffice from Sun Microsystems, for example).
/proc = a virtual directory providing stats on file system, processor, and memory etc.
/sbin = houses system binaries (such as reboot, halt, route, and fdisk) accessible to root only.
/tmp = a location for storing temporary data.
/usr = houses applications that are accessible to ALL users on the system such as a Web browser and gzip.
/var = provides space for housing data that changes frequently such as print/mail jobs and log files.

Cheers!

Linux System Manuals

2009-07-19T15:03:00.000-07:00

Linux manuals are typically stored in 9 structures -- man1 through man9. The manpath command shows the locations where manuals are located. Typically, manuals are stored under /usr/share/man.

Here is a brief description of what goes into each man structure:

man1 = user tools manuals
man2 = system calls manuals
man3 = c calls manuals
man4 = device driver manuals
man5 = configuration file manuals
man6 = game manuals
man7 = installation package manuals
man8 = system tools manuals
man9 = kernel manuals

The syntax of the command is:
# man

If a command is available in multiple manuals, you can specify which manual you want to look at by specifying the manual number like this:
# man [1-9]

Cheers!

Installing Broadcom Netlink Fast Ethernet Driver For Linux On Vostro 1400

2008-01-01T10:58:00.000-08:00

After I bought my new Vostro 1400 Notebook a few months ago, I installed CentOS 5 (read RHEL 5) on it and have been dual-booting it with Microsoft Windows Vista ever since. One big issue with installation of CentOS on Vostro 1400 is that the Broadcom Netlink Fast Ethernet card is not detected (the Intel ipw3945 Wireless card is also not recognized but details on that in a future post).

Research suggests that if you're running the latest 2.6 kernel with the tg3 module installed, your device should be recognized and working. Things are not that simple, though. The issue is that the tg3 module installed with the kernel 2.6 is old and doesn't recognize the ethernet card. All you need to do is install the latest version of the module (3.81c as of this writing), reboot your box, and Voila!

Here are step by step instructions for installing the Broadcom Netlink Fast Ethernet card on a Vostro 1400 Notebook running CentOS 5 (or RHEL 5 for that matter):

Note: You must have Internet connectivity to follow some of the steps specified below.

Fire up your browser and navigate to Broadcom Ethernet NIC NetLink Driver Downloads page. Download the driver titled Linux (tg3).
Open a terminal, login as root using the su command, and navigate to the directory where you downloaded the driver archive linux-3.81c.zip (latest as of this post).
Unzip the downloaded archive:
# unzip linux-3.81c.zip
Navigate to the directory just unzipped:
# cd Server/Linux/Driver/.
Untar the arvhice tg3-3.81c.tar.gz:
# tar -xvzf tg3-3.81c.tar.gz
Navigate to the untarred directory:
# cd tg3-3.81c
Make the kernel module:
# make
The make process should output various binaries including tg3.ko which is the kernel module we are interested in. So, install the module:
# make install
Update kernel module dependencies:
# /sbin/depmod -a
Finally, reboot your machine.

Once your machine reboots, the Broadcom Netlink Fast Ethernet card should be available for you to configure. Configure the card using ifconfig or the Network Configuration GUI tool available through the System > Administration > Network menu, and enjoy using your
Broadcom Netlink Fast Ethernet card under Linux on your Vostro 1400 box!

A Day Out

2007-12-08T19:11:00.000-08:00

Macrosoft's team working at Autoweb, which I am a key member of, went out on an all day event a week back. We ate lunch at Tawa Restaurant in Dearborn. After that we went to Henry Ford Museum. However, because we got so late, the museum was closed. So we went to Fairlane Shopping Mall where we did a bit of shopping. Then we watched the movie Beowulf in Star Theater. Finally, we had our dinner at China Town Restaurant.

That day it snowed very heavily; so much that on the way back to home I had to drive no more than 40 MPH while usually I do no less than 75 MPH!

mplayer Skins Problem

2007-09-08T16:53:00.000-07:00

On CentOS Linux (and may be others), the easiest way to enable mp3, rm, and other proprietary media support is to install mplayer (you'll need to have the rpmforge repo configured):

# yum install mplayer

I don't like the default skin though. So I installed the mplayer skins package:

# yum install mplayer-skins

After installing the skins package, I opened the mplayer skin browser and bang! No skin except the default one was available!! Research revealed that it's a small, but annoying, bug in the skins package which installs skins in a wrong directory. Check out details of this bug here.

Luckily, the resolution is very simple. All you need to do is execute this:

# cd /usr/share/mplayer/skins/
# ln -s ../Skin/* .

After executing these commands, restart mplayer, launch the skin browser and it should have a fairly long list of available skins listed. That's it!

Cheers.

Weird Shell Script Errors

2007-09-01T22:10:00.000-07:00

Ever ran into weird errors when trying to run a shell script which you believed was 100% correct but the shell won't accept it? I did! Well, it turned out that the bash shell (may be others as well) expects a script file to be saved with UNIX end of line format *not* Windows end of line format.

I am using CentOS 5 with KDE. I use the Kate editor for scripting and programming. Until I started getting insane errors whenever I would try to execute any of my recently developed shell scripts, Kate was configured to save files with Windows end of line format. Now that I changed the end of line format back to UNIX, bash is happy as ever!

Cheers.

Working Out of a Client Location in MI

2007-08-24T17:00:00.000-07:00

This post is in fact a little bit late. It's been a month since I have relocated to Rochester Hills, MI. But now is a good time to share with you a little bit of info about what I am doing here.

After three weeks of stay in NJ, and enjoying a very calm and cool work environment at Macrosoft, I got placed at a client's (Autoweb) site in Rochester Hills, MI. They are developing the new generation of a well-known product for the automotive industry. The main function of this product is providing a platform for sharing intellectual property data and collaboration among first tier automotive vendors like General Motors and Ford, and their suppliers.

My role in the development of this project is v. critical. In a nutshell, I am developing a component that would enable the old product and its new incarnation to work together seamlessly. This component's ultimate goal is to gradually migrate all clients from the old product to the new one and to enable vendors and suppliers to continue IP collaboration in the mean time regardless of which product they are using.

As you can see, this component which I am developing for my client is v. important for the success of the new product. And because we have almost run out of time-to-deadline, we're working not only long hours but weekends as well. And it's been like this for a month, non-stop. Man, this tough schedule! I didn't have to work like that even in Pakistan.

This is for you Pakis who always complaint that we have but unorganized software development industry where every other project is a chaos. Trust me, it's no better here either. In fact, software development is such a challenging and complex task that if it's not exercised with due caution and expertise, you're almost guaranteed to end up with a project that is a nightmare to develop, maintain, and evolve.

Stay tuned for more updates.
Cheers.

Relocation to the USA

2007-06-26T07:08:00.000-07:00

On Wednesday, 20 June, 2007 I relocated to the USA to work for Macrosoft Inc., a US-based software development and consulting company with operations in many countries including Pakistan. I flew to the US via PIA. Their service was good and I would like to travel with them in future as well.

New Jersey is a very green and calm state. It's really an ideal place to live in. It's been a week since I'm living and working in NJ and I'm really lovin' it!

Mplayer Font Problem -- Resolved!

2007-06-02T12:16:00.000-07:00

When I started mplayer after installing it on my CentOS 5 box, it thrown an error at me reading

New_Face failed. Maybe the font path is wrong. Please supply the text font file (~/.mplayer/subfont.ttf

The player also failed to play any media files. Any attempt to play an audio or video would result in a player crash. This later problem was resolved by following the instructions in the mplayer FAQs. Please note that this problem is specific to Fedora, RHEL, and CentOS.

The earlier font problem caused no issues as the player was able to play any media files. However, as you would expect, it was very, very annoying. Research revealed that solving this problem was as simple as copying any font file to the ~/.mplayer/ directory and renaming it to subfont.ttf. You can see the fonts available on your system by navigating to fonts:/ in your file browser (nautilus or konqueror).

Scheduling Tasks in Linux - cron

2007-05-13T06:19:00.000-07:00

On Linux systems, you use the crontab tool to schedule tasks that you want carried out on a certain time. Scheduled tasks are called cron jobs. The Linux Task Scheduler (read cron) has several parts to it:

1. The crond Daemon
crond is the workhorse that carries out the real job: executing specified tasks at the specified time. It's a daemon that runs calmly in the background and checks crontab files once every minute and executes tasks whose scheduled time matches the current time.

2. The crontab Tool
crontab is the tool that you use for editing your crontab file.

3. The crontab File
crontab files contain scheduling information of tasks.

The following sections examine each component of the cron system in detail.

The crond Daemon
The crond daemon is responsible for executing tasks at their scheduled time. It gets started automatically by rc scripts at system startup. So, you don't need to mess with it. But there might be times when you want to interact directly with the daemon. Here are a few useful commands to interact with the crond daemon or any other daemon for that matter:

To query status of crond:
# service crond status

To stop crond:
# service crond stop

To start crond:
# service crond start

To stop and start crond in one fell swoop:
# service crond restart

Please note that the service command resides in /sbin directory. On some Linux distros, notably Fedora Core, RHEL, and CentOS, this directory is not referenced in the PATH environment variables. To work around this, you need either to modify the PATH variable in your ~/.bash_profile file to reference /sbin or you need to specify the full path to the service command, i.e., /sbin/service.

The crond daemon executes cron jobs using the /bin/sh shell. Therefore, any environment variables that are available in /bin/sh shell are also available to cron jobs when they are executing.

The crontab Tool
crontab files are not meant to be edited directly. They are edited using the crontab tool. An important thing to note is that crontab files are created on a per-user basis. Whether or not a user can create crontab files is determined by cron.allow and cron.deny files in the /etc directory. Here's how the crontab tool determines permissions:

If neither cron.allow nor cron.deny files exist, only the root user can create crontab files.
If only cron.allow exists, the crontab tool allows a user to create crontab files if and only if her name is listed explicitly in the file. Other users are denied to create crontab files.
If only cron.deny exists, the crontab tool doesn't allow a user to create crontab files if her name is listed explicitly in the file. Other users are allowed to create crontab files.
If both cron.allow and cron.deny exist, the crontab tool doesn't allow a user
to create crontab files unless her name is explicitly listed in the cron.allow file. Other users are denied to create crontab file irrespective of whether their name is listed in the cron.deny file or not.

By default, the crontab tool executes for the user who is issuing the command. If you want to execute a command for a particular user, you'll need to specify the -u option followed by the user name whose crontab file you want to manipulate.

To edit your crontab file:
$ crontab -e

It will open your crontab file in an editor (usually vi) specified by the VISUAL or EDITOR environment variables. You'll need to edit the crontab file and save it. The crontab tool will take care of placing the file in an appropriate location.

To list contents of your crontab file:
$ crontab -l

To remove your crontab file:
$ crontab -r

If you need to list contents of the crontab file of a particular user, say jdoe, you'll execute the following command:
# crontab -u jdoe -l

The crontab file
Here is the format of the crontab file:
Minute Hour Day Month Day-Of-Week Command-To-Execute

Please note that you specify time in military format. Day-Of-Week runs from 0 through 6 representing Sunday through Saturday. Here are some examples of cron jobs:

If you want to shutdown your system at 23:59 every day except Saturday and Sunday, here's how you do it:
# 59 23 * * 1,2,3,4,5 shutdown "now"

Please note that if you want to specify multiple values for a given parameter, you separate them using commas (,) just as I did to specify Monday through Friday using 1,2,3,4,5.

Here's a cron job that executes every 5 minutes and appends current time to a file in the home directory of user jdoe:

# */5 * * * * echo "cron job executed at " $(date) >> /home/ejan/cron.out

Please note how we specified */5 to specify "every five minutes" rather than entering 0,5,10,15,...,55. If you want to execute your cron job at a regular interval like every five minutes, you can simply specify it using */5.

I hope his article will get you up to speed with scheduling tasks in Linux.

Linux Administration for Beginners

2007-05-06T05:39:00.000-07:00

Want to learn Linux administration quickly, easily, and in a practical, hands-on way? If yes, then the book Linux Administration: A Beginner's Guide is for you. Go buy yourself a copy from Amazon.

I've seen no other book that is more easy to follow, complete, practical, and fun to read than this one. If you hear someone saying administering Linux is difficult, just ask them to have a copy of this text. And I bet they'll be recommending it to *every* head interested in learning to administer Linux.

Have fun reading the book!

Mounting Samba Shares

2007-05-04T13:17:00.000-07:00

Lately I went through great troubles just to mount a samba share on my CentOS 5 box. Research revealed that a lot of people were having the same issue (see, for example, http://www.linuxquestions.org/questions/showthread.php?t=453941) of inability to mount samba shares despite everywhere the suggested syntax was the same (as in various help manuals). After research on possible solutions, I've found some rather interesting points regarding mounting samba shares that follow:

The following syntax works perfect for mounting samba shares:

mount -t cifs -o user=username //servername/sharename mountpoint

Here are some very important points about how the mount utility works (or not works) when it comes to mounting samba shares:

1. For the -t parameter smbfs *won't* work. That value for the -t parameter has retired and no longer supported.

2. If you specify a user whose password is *blank*, you'll need to press enter when prompted for password. In this case, mount will fail with a permission denied error. So always specify a user who does have a password.

3. sharename *must* be the root of a share not a subdirectory of it. For example, if you have a server named myserver with the following shared directory structure,

RootShare
..... |
.......... SubDirOne
.......... SubDirTwo
............... |
..................... SubSubDirOne

then you can mount nothing but the directory RootShare. Trying to mount any of the sub-directories will result in the following error:

retrying with upper case share name
mount error 6 = No such device or address
Refer to the mount.cifs(8) manual page (e.g.man mount.cifs)

4. mountpoint should be an already existing directory.

I hope this makes up for a crystal clear solution to a *very* common problem. Try googling for it and you'll see it seems like everyone on this planet has gone through this problem