Commands by olorin (9)

  • Configures screen to always display the clock in the last line (has to be configured only once). After that you not only have got the possibility to detach sessions and run them in background, but also have got a nice clock permanently on your screen.

    echo 'hardstatus alwayslastline " %d-%m-%y %c:%s | %w"' >> $HOME/.screenrc; screen
    olorin · 2011-02-16 08:04:56 1
  • If you provide the option -t to the script command and redirect stderr into a file, the timing information on what is going on on the terminal, is also stored. You can replay the session via the scriptreplay command, where you can also provide a speedup factor (see the man page for details). Great for demonstration purposes ...

    script -t /tmp/mylog.out 2>/tmp/mylog.time; <do your work>; <CTRL-D>; scriptreplay /tmp/mylog.time /tmp/mylog.out
    olorin · 2011-01-19 07:16:30 15
  • As an alternative to using an additional grep -v grep you can use a simple regular expression in the search pattern (first letter is something out of the single letter list ;-)) to drop the grep command itself. Show Sample Output

    ps aux | grep [p]rocess-name
    olorin · 2009-08-13 05:44:45 16
  • Within /proc and /sys there are a lot of subdirectories, which carry pseudofiles with only one value as content. Instead of cat-ing all single files (which takes quite a time) or do a "cat *" (which makes it hard to find the filename/content relation), just grep recursively for . or use "grep . /blabla/*" (star instead of -r flag). For better readability you might also want to pipe the output to "column -t -s : ". Show Sample Output

    grep -r . /sys/class/net/eth0/statistics
    olorin · 2009-08-05 08:20:39 1
  • Get the name of the parent command. This might be helpful, if you need to react on that information. E. g. a script called directly via ssh has got sshd as parent, manually invoked the parent process will probably be bash

    ps -o comm= -p $(ps -o ppid= -p $$)
    olorin · 2009-08-03 07:41:21 3
  • If you are doing some tests which require reboots (e. g. startup skripts, kernel module parameters, ...), this is very time intensive, if you have got a hardware with a long pre-boot phase due to hardware checks. At this time, kexec can help, which only restarts the kernel with all related stuff. First the kernel to be started is loaded, then kexec -e jumps up to start it. Is as hard as a reboot -f, but several times faster (e. g. 1 Minute instead of 12 on some servers here). Show Sample Output

    /sbin/kexec -l /boot/$KERNEL --append="$KERNELPARAMTERS" --initrd=/boot/$INITRD; sync; /sbin/kexec -e
    olorin · 2009-08-03 07:36:49 10
  • Instead of calculating the offset and providing an offset option to mount, let lomount do the job for you by just providing the partition number you would like to loop mount.

    lomount -diskimage /path/to/your/backup.img -partition 1 /mnt/foo
    olorin · 2009-07-22 11:32:52 14
  • You got some results in two variables within your shell script and would like to find the differences? Changes in process lists, reworked file contents, ... . No need to write to temporary files. You can use all the diff parameters you'll need. Maybe anything like $ grep "^>" is helpful afterwards.

    diff <(echo "$a") <(echo "$b")
    olorin · 2009-07-15 07:26:23 6
  • Let dd use direct I/O to write directly to the disk without any caching. You'll encounter very different results with different block sizes (try with 1k, 4k, 1M, ... and appropriate count values).

    time dd if=/dev/zero of=blah.out oflag=direct bs=256M count=1
    olorin · 2009-07-15 07:17:32 1

What's this? is the place to record those command-line gems that you return to again and again. That way others can gain from your CLI wisdom and you from theirs too. All commands can be commented on, discussed and voted up or down.

Share Your Commands

Check These Out

Prints total line count contribution per user for an SVN repository
I'm working in a group project currently and annoyed at the lack of output by my teammates. Wanting hard metrics of how awesome I am and how awesome they aren't, I wrote this command up. It will print a full repository listing of all files, remove the directories which confuse blame, run svn blame on each individual file, and tally the resulting line counts. It seems quite slow, depending on your repository location, because blame must hit the server for each individual file. You can remove the -R on the first part to print out the tallies for just the current directory.

Write a bootable Linux .iso file directly to a USB-stick
Writes hybrid ISO directly to USB stick; replace /dev/sdb with USB device in question and the ISO image link with the link of your choice

Compare two directory trees.
This uses Bash's "process substitution" feature to compare (using diff) the output of two different process pipelines.

command line Google I'm Feeling Lucky
opens the Google I'm Feeling Lucky result in lynx, the command line browser

Making scripts runs on backgourd and logging output
Save all output to a log.

cd into another dir to run a one-liner, but implicitly drop back to your $OLD_PWD after
Obviously the example given is necessarily simple, but this command not only saves time on the command line (saves you using "cd -" or, worse, having to type a fully qualified path if your command cd's more than once), but is vital in scripts, where I've found the behaviour of "cd -" to be a little broken at times.

ps a process keeping the header info so you know what the columns of numbers mean!
USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND root 1828 0.0 0.0 5396 476 ? Ss 2008 0:00 /usr/sbin/sshd

How to secure delete a file
Instead, install apt-get install secure-delete and you can use: -- srm to delete file and directory on hard disk -- smem to delete file in RAM -- sfill to delete "free space" on hard disk -- sswap to delete all data from swap

Prefix command output with duration for each line
Prefixes each output line with time between it and the previous one.

Add line number count as C-style comments
I often find the need to number enumerations and other lists when programming. With this command, create a new file called 'inputfile' with the text you want to number. Paste the contents of 'outputfile' back into your source file and fix the tabbing if necessary. You can also change this to output hex numbering by changing the "%02d" to "%02x". If you need to start at 0 replace "NR" with "NR-1". I adapted this from

Stay in the loop…

Follow the Tweets.

Every new command is wrapped in a tweet and posted to Twitter. Following the stream is a great way of staying abreast of the latest commands. For the more discerning, there are Twitter accounts for commands that get a minimum of 3 and 10 votes - that way only the great commands get tweeted.


Subscribe to the feeds.

Use your favourite RSS aggregator to stay in touch with the latest commands. There are feeds mirroring the 3 Twitter streams as well as for virtually every other subset (users, tags, functions,…):

Subscribe to the feed for: