Basically you enter openssh’s escape sequence ~ (only recognized immediately after newline), followed by Shift + C. This should give you a ssh> prompt. After that, you can use the port forwarding command option syntax without the leading dash.

1

user@sshhost $ ~C
ssh> L8080:localhost:8080
Forwarding port.

user@sshhost $

You can also cancel an existing port forwarding.

1

user@sshhost $ ~C
ssh> KL8080
Canceled forwarding.

user@sshhost $

If you are connected through multiple nested SSH connections, you can choose to modify the next deeper nested connection by adding one more ~ escape character.

There are additional useful escape sequences (for example to terminate hanging connections by ~.).

1

user@sshhost $ ~?
Supported escape sequences:
 ~.   - terminate connection (and any multiplexed sessions)
 ~B   - send a BREAK to the remote system
 ~C   - open a command line
 ~R   - request rekey
 ~V/v - decrease/increase verbosity (LogLevel)
 ~^Z  - suspend ssh
 ~#   - list forwarded connections
 ~&   - background ssh (when waiting for connections to terminate)
 ~?   - this message
 ~~   - send the escape character by typing it twice
(Note that escapes are only recognized immediately after newline.)

Cheers,
visit

I created a Dockerfile, which creates a custom container based on the official Tomcat 8 image. Additionally it adds the official sample application provided on apache.org.

1
2
3
4
5
6
7
8
9

FROM tomcat:8-jre8
MAINTAINER visit

ENV TOMCAT_VERSION 8.0
RUN curl -SL \
  -o /usr/local/tomcat/webapps/sample.war \
  "https://tomcat.apache.org/tomcat-${TOMCAT_VERSION}-doc/appdev/sample/sample.war"

CMD ["catalina.sh", "run"]

To make this Dockerfile available to all workers in my cluster, I uploaded it to etcd. The path is completely free to choose.

1

$ etcdctl set /units/tomcat8/Dockerfile < ./Dockerfile

Now I could create a service file for fleet. The instance notation (@) and specifiers (%p, %i) allow to start as many instances as I like, all from one single service file. For details on that review man 5 systemd.service and man 5 systemd.units. I also specified role=worker in the X-Fleet section, to let fleet automatically choose a suitable instance in my cluster.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19

[Unit]
Description=Tomcat 8 servlet container
After=docker.service
Requires=docker.service

[Service]
EnvironmentFile=/etc/environment
ExecStartPre=-/usr/bin/docker kill %p-%i
ExecStartPre=-/usr/bin/docker rm %p-%i
ExecStartPre=/usr/bin/mkdir -p /tmp/%p.%i
ExecStartPre=/usr/bin/bash -c "etcdctl get /units/%p/Dockerfile > /tmp/%p.%i/Dockerfile"
ExecStart=/usr/bin/bash -c "docker run -rm --name %p-%i -p 8080:8080 $(docker build /tmp/%p.%i | awk '/^Successfully built/ {print $3}')"
ExecStop=/usr/bin/docker stop %p-%i
Restart=always
TimeoutStartSec=0

[X-Fleet]
Conflicts=%p@*.service
MachineMetadata=role=worker

Last not least, I registered the service in fleet and started some instances.

1
2
3

$ fleetctl submit tomcat8@.service
$ fleetctl start tomcat8@1
$ fleetctl start tomcat8@2

Cheers,
visit

The first droplet is my etcd machine. Basically it runs a single etcd service, which is used as backend for fleet on all machines of my cluster. It was created with the following user-data:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34

#cloud-config

coreos:
  etcd:
    addr: $public_ipv4:4001
    peer-addr: $public_ipv4:7001
  fleet:
    public-ip: $public_ipv4
    metadata: "disk=ssd,platform=digitalocean,role=services"
  units:
    - name: etcd.service
      command: start
    - name: fleet.service
      command: start
    - name: docker-tcp.socket
      command: start
      enable: true
      content: |
        [Unit]
        Description=Docker Socket for the API

        [Socket]
        ListenStream=$public_ipv4:2375
        BindIPv6Only=both
        Service=docker.service

        [Install]
        WantedBy=sockets.target
write_files:
  - path: /etc/profile.d/aliases.sh
    permissions: 0644
    owner: root
    content: |
      alias ll='ls -l'

The exposed docker socket is not relevant for our cluster to function. I just used it to test docker images before deploying them to the whole cluster.

After the etcd droplet is running, fleetctl shows one machine with role=service as metadata. The metadata is used by fleet to decide, where clustered services should be started. I use role to distinguish between the one services machine and all other worker machines.

1
2
3

$ fleetctl list-machines
MACHINE		IP		METADATA
cfb065fa...	46.101.xxx.xxx	disk=ssd,platform=digitalocean,role=services

To create the worker machines I need to know the IP of the etcd droplet just created and insert it in the user-data of the workers. With that, workers can be created as many as I like, using the following user-data:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44

#cloud-config

coreos:
  fleet:
    public-ip: $public_ipv4
    metadata: "disk=ssd,platform=digitalocean,role=worker"
    etcd_servers: "http://46.101.xxx.xxx:4001"
  units:
    - name: etcd.service
      mask: true
    - name: fleet.service
      command: start
    - name: setup-etcdctl-env.service
      command: start
      content: |
          [Unit]
          Description=Setup etcdctl env vars
          Wants=coreos-setup-environment.service
          After=coreos-setup-environment.service

          [Service]
          Type=oneshot
          RemainAfterExit=true
          ExecStart=/bin/bash -c "source /etc/profile.d/etcdctl.sh; env | grep -E 'ETCDCTL|FLEETCTL|INSTANCE_ID' >>  /etc/environment"
          [Install]
          WantedBy=multi-user.target
write_files:
  - path: /etc/profile.d/etcdctl.sh
    permissions: 0644
    owner: core
    content: |
      export ETCDCTL_PEERS="http://46.101.xxx.xxx:4001"
  - path: /etc/profile.d/fleetctl.sh
    permissions: 0644
    owner: core
    content: |
      export FLEETCTL_ENDPOINT=unix:///var/run/fleet.sock
      export FLEETCTL_EXPERIMENTAL_API=true
write_files:
  - path: /etc/profile.d/aliases.sh
    permissions: 0644
    owner: root
    content: |
      alias ll='ls -l'

Instead of running etcd on every worker, I configured fleet to use the one from my etcd droplet. The setup-etcdctl-env.service makes it possible to use the etcdctl command from within fleet unit files, which I’ll describe in the next post.

To do failover testing with fleet I needed at least two workers.

1
2
3
4
5

$ fleetctl list-machines
MACHINE		IP		METADATA
9c4f0453...	46.101.xxx.xxx	disk=ssd,platform=digitalocean,role=worker
c0a4dd67...	46.101.xxx.xxx	disk=ssd,platform=digitalocean,role=worker
cfb065fa...	46.101.xxx.xxx	disk=ssd,platform=digitalocean,role=services

I can access each of them by:

1
2
3

$ fleetctl ssh cf
$ fleetctl ssh c0
$ fleetctl ssh 9

Cheers,
visit

First, I wanted a nice way to integrate DigitalOcean’s API into my desktop environment. It turns out, there is a Python2 based digitalocean-indicator system tray application around, which can be installed from the AUR, if you’re an Arch Linux user like me.

1

$ pacaur -Sa python2-digitalocean digitalocean-indicator

After that I, created a DigitalOcean account to test it. It work flawlessly and has a nice minimalist interface:

At this point I was able to start and stop containers from my desktop, but that’s not enough for a really good user experience.

As some of you may know, there is a lot of work done via SSH terminals if you use a cloud provider like DigitalOcean. Even more if you use CoreOS’s fleet. You usually control your container instances, system services and cluster components all via SSH. You should try to do as much of this, as possible from your local machine.

It turns out, you can manage a whole CoreOS cluster, by just knowing a single IP address of an etcd cluster node (which we can get from our system tray app).

First I needed to install the client applications (etcdctl, docker and fleetctl), which are part of their respective server packages:

1

$ pacaur -Sa docker etcd fleet

Each of these is able to work against a remote service, which can be specified via command-line argument or environment variable:

1
2
3
4

export DOCKER_HOST="tcp://46.101.xxx.xxx:2375"
export ETCDCTL_PEERS="http://46.101.xxx.xxx:4001"
export FLEETCTL_ENDPOINT="http://46.101.xxx.xxx:4001"
export FLEETCTL_STRICT_HOST_KEY_CHECKING="false"

FLEETCTL_STRICT_HOST_KEY_CHECKING makes fleetctl much more usable. Every time you re-image one of your CoreOS hosts you may get a ssh hostkey error when you try to use fleetctl. This variable disables this error only for fleetctl.

Now I can easily control my cluster from the local command-line.

1
2
3
4
5
6
7

$ etcdctl ls /
/coreos.com
$ docker images
REPOSITORY          TAG                 IMAGE ID            CREATED
$ fleetctl list-machines
MACHINE     IP      METADATA
cfb065fa... 46.101.xxx.xxx  disk=ssd,platform=digitalocean,role=services

Next: Run your own CoreOS cluster on DigitalOcean

Cheers,
visit

Some months ago I was searching for a nerd-style command-line based presentation tool and stumble across vtmc. I really liked the demo presentation, and the color fading feature blew me away. But as I didn’t like the concept of running node.js in a non client-server architecture, I thought about writing a similar tool without dependencies to an interpreter.

The second project which came to my knowledge lately was remarkable. A markdown editor for Linux written in python, which I really like. The syntax used in vtmc is quiet hard to memorize for a lazy guy like me, but markdown is used everywhere (in wikis, blogs, on github, etc.). So the idea evolved to combine the advantages of vtmc and the markdown language.

I always wanted to teach me some C programming. And this was finally a project, which was challenging enough to do so. So in the last weeks I went on to implement it. I had a really great time, and it was totally fun to implement such a project in a language so close to the hardware.

Now, after 5 weeks of spare time development, I finally uploaded a almost feature-complete (at least from my perspective) version to GitHub. Everything is completely written from scratch (beside ncurses). As the main purpose was to learn C programming, I added a lot of comments to the code. Maybe you find this helpful if you are in a similar position like me.

Now I need some beta testers, some ideas and helping hands to improve it. That’s why I would like to ask, if you are interested in joining my little project. Maybe you would like to download it and try? Or maybe check the code and / or improve it?

Hopefully I’m not the only nerd who likes it. :D

Cheers,
visit

1
2
3
4
5
6
7
8
9
10

# bind reverse-history-search to Ctrl+r
#          search: Ctrl+r > search-word > Enter
#     next result: n
# previous result: Shift+n
#
# be carefull while copying escape characters
# original command: trap '[[ ${.sh.edchar} == "^R" ]] && printf "^[[2K\r%s " "(reverse-history-search):" && .sh.edchar="^[/"' KEYBD
# ^R and ^[ are control characters
# insert them with Ctrl+V -> Ctrl+R and Ctrl+V -> Esc in vi edit mode
trap '[[ ${.sh.edchar} == "^R" ]] && printf "^[[2K\r%s " "(reverse-history-search):" && .sh.edchar="^[/"' KEYBD

Hope it helps,
visit

This looked really neat, but I’m to lazy to configure an entire firewall just to open/close one port, beside I already have an firewall on my router. The second thing is I want to connect to my server without port knocking when I’m in my home network.

So here is my solution for that. Instead of opening and closing a port with knockd, I just enable and disable a local port forwarding via iptables.

1
2
3
4
5
6
7
8
9
10
11

[options]
    UseSyslog
    Interface = enp2s0

[opencloseSSH]
    sequence      = 4443:tcp,4442:tcp,4444:tcp,4441:tcp,4445:tcp
    seq_timeout   = 15
    tcpflags      = syn
    start_command = /usr/sbin/iptables -t nat -A PREROUTING -s %IP% -p tcp --dport 4440 -j REDIRECT --to-port 22
    cmd_timeout   = 10
    stop_command  = /usr/sbin/iptables -t nat -D PREROUTING -s %IP% -p tcp --dport 4440 -j REDIRECT --to-port 22

As long as your default policy for the INPUT chain is ACCEPT this doesn’t even require an additional rules for RELATED,ESTABLISHED connections.

On the router you just need to forward the port range used for knocking and of cause port 22 (to 4440 on your home server) for SSH ;)

Hope this helps,
visit

Create a multidimensional hash variable:

1
2
3
4
5
6
7
8
9
10
11

$ typeset -A a=([b]='')
$ typeset -A a[b]=([c]='')
$ typeset -A a[b][c]=([d]='e')
$ printf "%B\n" a
(
        [b]=(
                        [c]=(
                                        [d]=e
                        )
        )
)

Extend it with additional values:

1
2
3
4
5
6
7
8
9
10

$ typeset -A a[b][c]+=([f]='g')
$ printf "%B\n" a
(
        [b]=(
                        [c]=(
                                        [d]=e
                                        [f]=g
                        )
        )
)

Work with the dimensions to manipulate output:

1
2
3
4
5
6
7
8
9
10

$ echo ${!a[*]}
b
$ echo ${!a[b][*]}
c
$ echo ${!a[b][c][*]}
d f
$ echo ${a[b][c][d]}
e
$ echo ${a[b][c][f]}
g

Hope it helps,
visit

Extract audio only

1

ffmpeg -i inputfile.m4v -vn -acodec copy outputfile.aac

Extract video only

1

ffmpeg -i inputfile.m4v -an -vcodec copy outputfile.h246

Merge video and audio

1

ffmpeg -i audio.aac -i video.h246 -acodec copy -vcodec copy output.m4v

Hope it helps,
visit

Vim Novice Video Tutorials
Vim Intermediate Video Tutorials
Vim Advanced Video Tutorials
Vim Plugin Showcase

Hope it helps,
visit

1

alias calc='bc -l <<<'

This one lets you calculate directly in the bash terminal like that: calc 1+1
If you wanna use brackets you must enclose the expression in quotes: calc "(4+6)/2"

1

alias ports='lsof -i -P'

This one shows all network connections to your host. It displays more information than netstat -anp and formats it in a nice way.

1

alias psu='ps -fHU $USER'

This one shows the processes of the current user in hierarchical order. Nice to see whats going on.

1

alias vip='sudo su'

I use this as long as I can remember for switching to root.

Hope it helps,
visit

You can find the source on GitHub.

It’s depending on csync >0.50.8. and pyinotify (tested with python 2.7.3).

It creates a small config file in ~/.config/ocpy/

The logfile and pidfile are hardcoded to /tmp/ocpy.[log|pid] in case you
wanna place them somewhere else.

Hope this helps,
visit

So here just a quick note about how to convert an entire album to mp3.

1

find . -name \*.flac -exec ffmpeg -i "{}" -ab 128k -ac 2 -ar 44100 "{}.mp3" \; && rm *.flac

Hope it helps,
visit

1

git push origin newfeature

Deleting remote branches is easier that creating them.

1

git push origin :newfeature

Translated to english from here.

Since this is a little tricky I decided to post this here:

First you have to add the parent repository as a remote repository of your fork.

1

git remote add <alias> <url>

Then pull the changes with the –rebase option which prevents git from creating a commit for the merge.

1

git pull --rebase <alias> <branch>

In case of conflicts:

Use this to find them:

1

git diff

Open the appropriate file and search for ‘<<<’ to find the lines to fix.

Run this to mark fixed conflict files as resolved:

1

git add <file>

Then complete the rebase with:

1

git rebase --continue

Hope this helps,
visit

After a big memory upgrade I don’t turn off my linux box any more. And since there is enough memory to run a passive virtual machine, I decided to install iTunes in VirtualBox and run it in background.

So here is a quick walk-through:

First you have to install VirtualBox from virtualbox.org. Since I’m on Linux Mint 12, I chose the Ubuntu 11.10 .deb package and installed it as followed.

1

sudo dpkg -i path/to/debfile.deb

Apple devices need USB2.0 support. To get this working with VirtualBox you have to download the Oracle VM VirtualBox Extension Pack from oracle.com. After installing VirtualBox open the global settings dialog and add the downloaded file under extension packs.

Now create a new virtual machine. Make sure you pick “Bridge network” in the network section to get Wi-Fi Sync working and enable USB2.0 support in the USB section.

At this point you can also add filters to the USB section to automatically forward Apple devices to the virtual machine on connect. You have to connect each device one after another and choose the Apple device from the “Add Filter From Device” button.

Since I want to use my existing iTunes library from my Windows machine and want to use the library for other services like Subsonic, I decided to store the library on the host system. To make it available in the virtual machine I created a shared folder to it in the settings too.

Now install your guest Windows system and the VirtualBox Guest Additions from the devices menu.

After that download and install iTunes from apple.com. If you want to use a existing iTunes library like me, uncheck the “start iTunes after installing” at the end of the installer.

Make sure your shared folder with the iTunes library is available and startup iTunes while holding down the [Shift]-key. This will bring up a dialog where you can choose your existing library.

Now connect your Apple device via USB and hopefully it will synchronize correctly.

Hope it helps,
visit

P.S.: Don’t use WinSCP to copy your iTunes library to linux since it can’t handle utf-8 special characters correctly. I used the cygwin port of rsync instead.

At the end I can say: It isn’t very hard to write a system tray application in python. The really challenging thing is to use threading for the csync process and create a settings dialog to configure your target instance of ownCloud.

After four weeks of after work coding here it is. My first little python application: ownCloudTray

Download:
Version 0.1.0 - ownCloudTray-0.1.0.tar.gz
Version 0.2.0 - Source moved to GitHub

Todos:

• Implementing a notification when files are synchronized from cloud to desktop.

Please comment below if you found issues or have suggestions how to make it better.

Hope it helps,
visit

I found a very helpfull post about compiling it for Ubuntu 9.10.

The dependencies changed a bit between csync version 0.44.0 and 0.50.0. So i added the full list for 0.50.0 below:

• cmake
• check
• log47
• sqlite3
• libiniparser
• libsmbclient
• libssh
• libneon
• doxygen

Here is the command to install all needed dependencies:

1

sudo apt-get install cmake check liblog4c3 liblog4c-dev libsqlite3-0 libsqlite3-dev libsmbclient-dev libssh-4 libssh-dev libneon27 libneon27-dev doxygen checkinstall

I downloaded the source of csync via git:

1

git clone git://git.csync.org/projects/csync.git

Now you can follow the instructions on gremu.net. You have to install libiniparser as described. Especially the last part about building a .deb file is very good.

Because csync could not find its shared libraries in /usr/local/lib, I changed all paths in csync/build/CMakeCache.txt from /usr/local to /usr before calling cmake.

Here the resulting csync_0.50.0-1_amd64.deb. This one only works with 64bit Linux!

To run csync with owncloud type something similar to this:

1

csync /home/user/owncloud owncloud://user:password@cloud.domain.tld/files/webdav.php/

Comment below if you found issues or have suggestions how to make it better.

Hope it helps,
visit

did you ever want to create a dialog for your favorite commandline tool or are you tired of typing the same 2 inch command every time again?

Then zenity can make your life much easier!

This tool comes with a set of configurable dialog boxes from simpe info-, warning- and error-boxes to complex file-selection, progress-bar or forms.

I’ve build a small wrapper to get rid of the commandline tool youtube-dl for example.

The wrapper pops up a input-box for the source url and a directory-selection dialog for the destionation folder. Make a desktop- or menu-shortcut to it and everything else will runs in the background.

Hope it helps,
visit