This section covers some very basic commands to efficiently work with a Linux system. They should be available on all standard installations. Although many modern Linux variants offer an advanced GUI, it is advisable to be able to control a Linux system from command line.
lslists files in current directory. Use
ls -alto list all files (including hidden = starting with .) in a long listing.
mkdircreates a directory. Specify the directory name after the command.
rmremoves files and directories. If directories need to be removed, the
-roption must be specified. To force operation, add
-f. The last parameters are the files and directories that are to be deleted. Make sure that there are no wrong directories listed! A call to
rm -rf /as root is not a good thing.
mvmoves a file, where path is relative to current directory: mv [path]filename [path]filname
filereveals what content type a file is. Use it to test if a file is text, image, binary or whatever.
rmcan be used in combination to delete files found by
find. e.g. this will recursively remove all files with extension
.txtin the current directory:
find . -name \*.txt -print0 | xargs -0 rm
rm $(find -name \*.txt)
sudostarts a process with super user (=root) capabilities. To start a root shell, use
sudo su. If
sudois not installed, use
suinstead. It requires not the current user's password but the root password.
lesshelps to look at what a file stores.
catreads a file and prints it to
stdout(the console if invoked as-is.) To read from STDIN, pass
-as the file to read. This can be used to create a file by entering its content in the console. Use
cat - > new_fileand paste the content to quickly create a new file with content. To end the file, press Ctrl+D, which produces the EOF-character.
locatelocates all files that contain a given string.
grepsearches input for matching lines. To search all files in the current directory (and sub directories), use
grep -r . -e “text”.
manshows the manual for some command. Sometimes
infoshows a more detailed documentation.
dmesgprints kernel messages.
wgetto download files from HTTP and FTP. The tool allows recursive downloading, i.e. also fetch outgoing links from some HTML page. To restrict the files downloaded to a certain type (by file suffix), use the
wget -r http://some.host/path/to/file.html -A pdf
downloads all PDFs referenced by the HTML file.
rsync. It allows recursive uploading, can be configured to upload changed or new files only, can optionally delete remote files that are no longer existing locally, and is very nice to your bandwith.
* This is a script to upload a complete directory recursively (local copy of website to web server).
The local directory is
gallery, this is also the place where the script resides. The remote username
user and should be replaced by an existing one. Also, the server
server.xx and the server directory have to be adapted.
The rsync options used here:
-a: keep a lot of attributes (owner, permissions etc.);
-r: recurse into subdirectories;
–update: only upload changed or added files;
–delete: delete remote files no
longer available locally; and finally:
–delay-updates: makes rsync postpone all changes until the run is finished, which makes a lot of sense when updating live websites.
If you comment out the second
myoption= line and uncomment the first one, the
–dry-run option lets you simulate what rsync would do without actually harming anything.
#! /bin/sh #comment out --dry-run to get it working #myoption=--dry-run # don't put spaces around the = myoption= #this resets the variable echo "--> gallery dir" rsync -razv --update --delete $myoption --delay-updates gallery email@example.com:/var/www/de.baz.foo/
The output of one command can be piped into the input of another command. This is done by the pipe operator
|. To search kernel messages for some string, use
dmesg | grep string.
Additionally, the STDOUT (1) and STDERR (2) streams can also be redirected. To redirect the STDERR output to STDOUT, use:
command 2>&1 …. This can be used in combination with piping.