Restore an etc configuration file from the original maintainer version on Debian

How to, Linux, , , , , , , ,
There are events that can not be stopped, for example when you delete a file by mistake under /etc.

Debian provides an elegant way to restore files of the maintainer’s version without touching your existing configuration.

Practical scenario

Let’s say I’ve removed the file /etc/apache2/sites-available/000-default.conf, here’s the procedure to recover it:

  • find the deb package containing the file 
    # dpkg -S /etc/apache2/sites-available/000-default.conf
  • reinstall the package with a specific option 
    # apt-get install --reinstall -o Dpkg::Options::="--force-confmiss" apache2
Reading package lists... Done
Building dependency tree       
Reading state information... Done
0 upgraded, 0 newly installed, 1 reinstalled, 0 to remove and 5 not upgraded.
Need to get 86,7 kB of archives.
After this operation, 0 B of additional disk space will be used.
Get:1 http://archive.ubuntu.com/ubuntu/ saucy-updates/main apache2 amd64 2.4.6-2ubuntu2.1 [86,7 kB]
Fetched 86,7 kB in 0s (264 kB/s)
(Reading database ... 245864 files and directories currently installed.)
Preparing to replace apache2 2.4.6-2ubuntu2.1 (using .../apache2_2.4.6-2ubuntu2.1_amd64.deb) ...
Unpacking replacement apache2 ...
Processing triggers for man-db ...
Processing triggers for ufw ...
Processing triggers for ureadahead ...
ureadahead will be reprofiled on next reboot
Setting up apache2 (2.4.6-2ubuntu2.1) ...

Configuration file `/etc/apache2/sites-available/000-default.conf', does not exist on system.
Installing new config file as you requested.
 * Restarting web server apache2
   ...done.

That’s all.

Practical scenario #2

There’s another interesting case when this procedure comes handy: if you changed a default etc file and you want to restore it. In that case you can simply delete it and use the same apt-get command:

# apt-get install --reinstall -o Dpkg::Options::="--force-confmiss" <package name>

References

Meet the “It’s All Text!” Firefox extension

Uncategorized, , , , , , , , , ,
Yes, I’m a vim addicted. I mean vim the text editor. I’m so addicted to it that I tend to use vim’s shortcuts even in LibreOffice documents, thunderbird e-mails, web forms etc. with awful results.

Then I found It’s All Text, a Firefox extension which saved my web experience when editing long textareas.

It’s really simple: it puts an overlay “edit” button beside each textarea which is visible only when overing it. When you click it, your preferred editor will come up filled of the text of the textarea. Now you can make your edits, close it and boom, the text is in the textarea.

Wonderful.

For me it’s a productivity boost.

I use it everywhere, like now, I’m writing this blog post in my preferred editor.

The nerd tip

There a deeper tip I use: I’m opening vim from a gnome-terminal.

I created a bash script under my $HOME/bin directory with this name editor-for-it-s-all-text.sh and which content is

#!/bin/bash
/usr/bin/gnome-terminal --full-screen -e "vim '$1'"

Then I configured “It’s All Text” to use that editor and now I use my beloved vim in a full screen terminal window.

Awesome.

Dockerfile: “no such file or directory” error using ADD

How to, Linux, , , , ,
Testing docker is really interesting, but sometimes it’s difficult to understand what’s wrong with some configuration.

A problem I found recently dealed with the ADD directive used in the Dockerfile. I was trying to start some services with supervisor but I got this error during the image build process

Step 19 : ADD supervisord.conf /etc/supervisor/conf.d/
2014/02/10 00:40:55 build: supervisord.conf: no such file or directory

The file was right there, in the same path of the Dockerfile, but docker couldn’t find it.

After a good read of the official documentation I learned the conxept of “build context”. When you are building an image, the source directory from which you are operating is the build context, but when you are building passing the Dockerfile from the standard input, there’s no build context!

So this is ok

$ docker build -t mydebian .

and this can’t work

$ docker build -t mydebian - < Dockerfile

Cache deb packages with apt-cacher-ng

How to, Linux, , , , , , , , ,
I’m playing with Docker and I’m constantly tuning my Dockerfile to install the needed packages.

This continuous refinement forces the build process to download the same deb packages every time and it becomes an actual waste of time and bandwidth.

Solution

apt-cacher-ng is a service which can be run on a Debian or Ubuntu host and will serve as a local cache for deb packages from Debian or Ubuntu repositories.

That’s what I need because I’m on an Ubuntu machine emulating a Debian container.

Host setup

The setup is really easy:

# aptitude install apt-cacher-ng

After the installation I changed the bind address of the service to keep it local (I don’t need LAN exposure), I added my local proxy to reach the internet and I disabled ReuseConnections (see Problems):

root@yoda:/etc/apt-cacher-ng# diff acng.conf acng.conf.old 
28d27
< BindAddress: localhost 172.17.42.1
35d33
< proxy: http://127.0.0.1:5865
322d319
< ReuseConnections: 0

Client setup

On the client side it’s necessary to force apt requests to use the proxy.
In the Dockerfile I added

RUN echo 'Acquire::http::Proxy "http://172.17.42.1:3142";' > /etc/apt/apt.conf.d/90-apt-cacher.conf

which creates the configuration file read by apt-get or aptitude when downloading packages.

Problems

The apt-cacher-ng server seems buggy at least when used in conjunction with a regular proxy. Sometimes I get errors like these on the client side

Err http://http.debian.net/debian/ wheezy/main netbase all 5.0
500  Invalid header

one possible solution is to perform a

# apt-get update

on the host machine (it seems to help). Another useful setting is

ReuseConnections: 0

as explained before.