========= Internals ========= Introduction ============ A ConPaaS application represents a collection of ConPaaS services working together. The application manager is a process that resides in the first VM that is created when the application is started and is in charge of managing the entire application. The application manager represents the single control point for the entire application. A ConPaaS service consists of three main entities: the service manager, the service agent, and the web frontend. The service manager is a component that supplements the application manager with service-specific functionality. Its role is to manage the service by providing supporting agents, maintaining a stable configuration at any time and by permanently monitoring the service’s performance. A service agent resides on each of the VMs that are started by the service. The agents are the ones doing all the work. To implement a new ConPaaS service, you must provide a new service manager, a new service agent and a new service frontend (we assume that each ConPaaS service can be mapped on the three entities architecture). To ease the process of adding a new ConPaaS service, we propose a framework which implements the common functionality of the ConPaaS services. So far, the framework provides abstractions for the IaaS layer (adding support for a new cloud provider should not require modifications in any ConPaaS service implementation) and it also provides abstractions for the HTTP communication (we assume that HTTP is the preferred protocol for the communication between the three entities). ConPaaS directory structure --------------------------- You can see below the directory structure of the ConPaaS software. The *core* folder under *src* contains the ConPaaS framework. Any service should make use of this code. It contains base classes for service managers and agents, and other useful code. A new service should be added in a new python module under the *ConPaaS/src/conpaas/services* folder: :: ConPaaS/ (conpaas/conpaas-services/) │── src │ │── conpaas │ │   │── core │ │   │   │── clouds │ │   │   │   │── base.py │ │   │   │   │── dummy.py │ │   │   │   │── ec2.py │ │   │   │   │── federation.py │ │   │   │   │── openstack.py │ │   │   │── agent.py │ │   │   │── callbacker.py │ │   │   │── expose.py │ │   │   │── file.py │ │   │   │── ganglia.py │ │   │   │── git.py │ │   │   │── https │ │   │   │── log.py │ │   │   │── manager.py │ │   │   │── misc.py │ │   │   │── node.py │ │   │   │── services.py │ │   │── services │ │   │── flink/ │ │   │── generic/ │ │   │── helloworld/ │ │   │── mysql/ │ │   │── webservers/ │ │   │── xtreemfs/ │ │── setup.py │── config │── contrib │── misc │── sbin │── scripts In the next paragraphs, we describe how to add the new ConPaaS service. Implementing a new ConPaaS service ================================== In this section, we describe how to implement a new ConPaaS service by providing an example which can be used as a starting point. The new service is called *helloworld* and will just generate helloworld strings. Thus, the manager will provide a method, called get\_helloworld which will ask all the agents to return a ’helloworld’ string (or another string chosen by the manager). We will start by implementing the agent. We will create a class, called HelloWorldAgent, which implements the required method - get\_helloworld, and put it in *conpaas/services/helloworld/agent/agent.py* (Note: make the directory structure as needed and providing empty \_\_init\_\_.py to make the directory be recognized as a module path). As you can see in |lst:helloworldagent|, this class uses some functionality provided in the conpaas.core package. The conpaas.core.expose module provides a python decorator (@expose) that can be used to expose the http methods that the agent server dispatches. By using this decorator, a dictionary containing methods for http requests GET, POST or UPLOAD is filled in behind the scenes. This dictionary is used by the built-in server in the conpaas.core package to dispatch the HTTP requests. The module conpaas.core.http contains some useful methods, like HttpJsonResponse and HttpErrorResponse that are used to respond to the HTTP request dispatched to the corresponding method. In this class, we also implemented a method called startup, which only prints a line of text in the agent's log file. This method could be used, for example, to make some initializations in the agent. .. |lst:helloworldagent| replace:: Listing 7 .. centered:: |lst:helloworldagent|: conpaas/services/helloworld/agent/agent.py .. literalinclude:: ../conpaas-services/src/conpaas/services/helloworld/agent/agent.py Let’s assume that the manager wants each agent to generate a different string. The agent should be informed about the string that it has to generate. To do this, we could either implement a method inside the agent, that will receive the required string, or specify this string in the configuration file with which the agent is started. We opted for the second method just to illustrate how a service could make use of the config files and also, maybe some service agents/managers need some information before having been started. Therefore, we will provide the *helloworld-agent.cfg* file (see |lst:helloworldcfg|) that will be concatenated to the default-manager.cfg file. It contains a variable ($STRING) which will be replaced by the manager. .. |lst:helloworldcfg| replace:: Listing 8 .. centered:: |lst:helloworldcfg|: ConPaaS/config/agent/helloworld-agent.cfg .. literalinclude:: ../conpaas-services/config/agent/helloworld-agent.cfg :language: cfg Now let’s implement an http client for this new agent server. See |lst:helloworldagentclient|. This client will be used by the manager as a wrapper to easily send requests to the agent. We used some useful methods from conpaas.core.http, to send json objects to the agent server. .. |lst:helloworldagentclient| replace:: Listing 9 .. centered:: |lst:helloworldagentclient|: conpaas/services/helloworld/agent/client.py .. literalinclude:: ../conpaas-services/src/conpaas/services/helloworld/agent/client.py Next, we will implement the service manager in the same manner: we will write the *HelloWorldManager* class and place it in the file *conpaas/services/helloworld/manager/manager.py*. (See |lst:helloworldmanagermanager|) A service manager supplements the application manager with service-specific functionality. It does so by overriding the methods inherited from the base manager class. These methods will be called by the application manager when the corresponding event occurs. For example, *on\_start* is called immediately after the service is started, *on\_add\_nodes* after additional nodes have been added to the service, *on\_remove\_nodes* after nodes have been removed, *on\_stop* after the service was stopped. .. |lst:helloworldmanagermanager| replace:: Listing 10 .. centered:: |lst:helloworldmanagermanager|: conpaas/services/helloworld/manager/manager.py .. literalinclude:: ../conpaas-services/src/conpaas/services/helloworld/manager/manager.py The last step is to register the new service to the conpaas core. One entry must be added to file *conpaas/core/services.py*, as it is indicated in |lst:helloworldservices|. Because the Java and PHP services use the same code for the agent, there is only one entry in the agent services, called *web* which is used by both webservices. .. |lst:helloworldservices| replace:: Listing 12 .. centered:: |lst:helloworldservices|: conpaas/core/services.py .. literalinclude:: ../conpaas-services/src/conpaas/core/services.py Integrating the new service with the frontend ============================================= So far, there is no easy way to add a new frontend service. Each service may require distinct graphical elements. In this section, we explain how to create the web frontend page for a service. Manager states -------------- As you have noticed in the Hello World manager implementation, we used some standard states, e.g. INIT, ADAPTING, etc. By calling the *get\_service\_info* function, the frontend knows in which state the manager is. Why do we need these standardized stated? As an example, if the manager is in the ADAPTING state, the frontend would know to draw a loading icon on the interface and keep polling the manager. Files to be modified -------------------- :: frontend │── www │── create.php │── lib │── service │── factory │── __init__.php Several lines of code must be added to the two files above for the new service to be recognized. If you look inside these files, you’ll see that knowing where to add the lines and what lines to add is self-explanatory. Files to be added ----------------- .. role:: red :: frontend │── www │── lib | │── service | | │── helloworld | | │── __init__.php | │── ui | │── instance | │── helloworld | │── __init__.php │── images │── helloworld.png .. _image-creation: Creating A ConPaaS Services VM Image ==================================== Various services require certain packages and configurations to be present in the VM image. ConPaaS provides facilities for creating specialized VM images that contain these dependencies. Furthermore, for the convenience of users, there are prebuilt images that contain the dependencies for *all* available services. If you intend to use these images and do not need a specialized VM image, then you can skip this section. Configuring your VM image ------------------------- The configuration file for customizing your VM image is located at ``conpaas-services/scripts/create_vm/create-img-script.cfg``. In the **CUSTOMIZABLE** section of the configuration file, you can define whether you plan to run ConPaaS on Amazon EC2 or OpenStack. Depending on the virtualization technology that your target cloud uses, you should choose either KVM or Xen for the hypervisor. Note that for Amazon EC2 this variable needs to be set to Xen. Please do not make the recommended size for the image file smaller than the default. The *optimize* flag enables certain optimizations to reduce the necessary packages and disk size. These optimizations allow for smaller VM images and faster VM startup. In the **SERVICES** section of the configuration file, you have the opportunity to disable any service that you do not need in your VM image. If a service is disabled, its package dependencies are not installed in the VM image. Paired with the *optimize* flag, the end result will be a minimal VM image that runs only what you need. Note that te configuration file contains also a **NUTSHELL** section. The settings in this section are explained in details in :ref:`conpaas-in-a-nutshell`. However, in order to generate a regular customized VM image, make sure that both *container* and *nutshell* flags in this section are set to *false*. Once you are done with the configuration, you should run this command in the ``create_vm`` directory:: $ python create-img-script.py This program generates a script file named ``create-img-conpaas.sh``. This script is based on your specific configurations. Creating your VM image ---------------------- To create the image you can execute ``create-img-conpaas.sh`` in any 64-bit Debian or Ubuntu machine. Please note that you will need to have root privileges on such a system. In case you do not have root access to a Debian or Ubuntu machine please consider installing a virtual machine using your favorite virtualization technology, or running a Debian/Ubuntu instance in the cloud. #. Make sure your system has the following executables installed (they are usually located in ``/sbin`` or ``/usr/sbin``, so make sure these directories are in your ``$PATH``): **dd parted losetup kpartx mkfs.ext3 tune2fs mount debootstrap chroot umount grub-install** #. It is particularly important that you use Grub version 2. To install it:: sudo apt-get install grub2 #. Execute ``create-img-conpaas.sh`` as root. The last step can take a very long time. If all goes well, the final VM image is stored as ``conpaas.img``. This file is later registered to your target IaaS cloud as your ConPaaS services image. If things go wrong ------------------ Note that if anything fails during the image file creation, the script will stop and it will try to revert any change it has done. However, it might not always reset your system to its original state. To undo everything the script has done, follow these instructions: #. The image has been mounted as a separate file system. Find the mounted directory using the ``df -h`` command. The directory should be in the form of ``/tmp/tmp.X``. #. There may be a ``dev`` and a ``proc`` directories mounted inside it. Unmount everything using:: sudo umount /tmp/tmp.X/dev/pts /tmp/tmp.X/dev /tmp/tmp.X/proc /tmp/tmp.X #. Find which loop device you are using:: sudo losetup -a #. Remove the device mapping:: sudo kpartx -d /dev/loopX #. Remove the binding of the loop device:: sudo losetup -d /dev/loopX #. Delete the image file #. Your system should be back to its original state. .. _creating-a-nutshell: Creating a Nutshell image ========================= Starting with the release 1.4.1, ConPaaS is shipped together with a VirtualBox appliance containing the Nutshell VM image. This section explains how to create a similar image that can be deployed on a different virtualization technology (such as the other clouds supported by ConPaaS). The next section describes the procedure for recreating the VirtualBox image. If you are interested only in installing the standard VirtualBox image that is shipped with ConPaaS, you may skip this chapter entirely and only read the installation guide available here: :ref:`conpaas-in-a-nutshell`. The procedure for creating a Nutshell image is very similar to the one for creating a standard customized image described in section :ref:`image-creation`. However, there are a few settings in the configuration file which need to be considered. Most importantly, there are two flags in the **Nutshell** section of the configuration file, *nutshell* and *container* which control the kind of image that is going to be generated. Since these two flags can take either value *true* of *false*, we distinguish four cases: #. *nutshell = false*, *container = false*: In this case, a standard ConPaaS VM image is generated and the nutshell configurations are not taken into consideration. This is the default configuration which should be used when ConPaaS is deployed on a standard cloud. #. *nutshell = false*, *container = true*: In this case, the user indicates that the image that will be generated will be a LXC container image. This image is similar to a standard VM one, but it does not contain a kernel installation. #. *nutshell = true*, *container = false*. In this case, a Nutshell image is generated and a standard ConPaaS VM image will be embedded in it. This configuration should be used for deploying ConPaaS in nested standard VMs within a single VM. #. *nutshell = true*, *container = true*. Similar to the previous case, a Nutshell image is generated but this time a container image is embedded in it instead of a VM one. Therefore, in order to generate a Nutshell based on LXC containers, make sure to set these flags to this configuration. This is the default configuration for our distribution of the Nutshell. Another important setting for generating the Nutshell image is also the path to a directory containing the ConPaaS tarballs (cps*.tar.gz files). The rest of the settings specify the distro and kernel versions that the Nutshell VM would have. In order to run the image generating script, the procedure is almost the same as for a standard image. From the ``create_vm`` directory run:: $ python create-img-script.py $ sudo ./create-img-nutshell.sh Note that if the *nutshell* flag is enabled the generated script file is called ``create-img-nutshell.sh``. Otherwise, the generated script file is called ``create-img-conpaas.sh`` as indicated previously. Creating a Nutshell image for VirtualBox ---------------------------------------- As mentioned earlier the Nutshell VM can also run on VirtualBox. In order to generate a Nutshell image compatible with VirtualBox, you have to set the *cloud* value to *vbox* in the **Customizable** section of the configuration file. The rest of the procedure is the same as for other clouds. The result of the image generation script would be a ``nutshell.vdi`` image file which can be used as a virtual hard drive when creating a new appliance on VirtualBox. The procedure for creating a new appliance on VirtualBox is quite standard: #. Name and OS: You choose a custom name for the appliance but use *Linux* and *Ubuntu (64 bit)* for the type and version. #. Memory size: Since the Nutshell runs a significant number of services and also requires some memory for the containers, we suggest to choose at least 4 GB of RAM. #. Hard drive: Select "User an existing virtual hard drive file", browse to the location of the ``nutshell.vdi`` file generated earlier and press *create*. .. _preinstall-app-in-conpaas-image: Preinstalling an application into a ConPaaS Services Image ========================================================== A ConPaaS Services Image contains all the necessary components needed in order to run the ConPaaS services. For deploying arbitrary applications using ConPaaS, :ref:`the-generic-service` provides a mechanism to install and run the application, along with its dependencies. The installation, however, has to happen during the initialization of every new node that is started, for example in the ``init.sh`` script of the Generic Service. If installing the application with its dependencies takes a long time or, in general, is not desired to happen during every deployment of a new node, another option is available: preinstalling the application inside the ConPaaS Services Image. The current section describes this process. #. Download a ConPaaS Services Image appropriate for your computer architecture and virtualization technology. Here are the download links for the latest images: **ConPaaS VM image for Amazon EC2 (x86_64):** | http://www.conpaas.eu/dl/conpaas-2.1.0-amazon.img.tar.gz | MD5: 2d2af415a29c8413b898eacfcc2f1343 | size: 541 MB **ConPaaS VM image for OpenStack with KVM (x86_64):** | http://www.conpaas.eu/dl/conpaas-2.1.0-openstack-kvm.img.tar.gz | MD5: 02ea7ef89ff81bf03668003a8d2feac6 | size: 541 MB **ConPaaS VM image for OpenStack with LXC (x86_64):** | http://www.conpaas.eu/dl/conpaas-2.1.0-openstack-lxc.img.tar.gz | MD5: f04e9dcc3059c5d10f599e72243055d3 | size: 509 MB **ConPaaS VM image for OpenStack with LXC for the Raspberry Pi (arm):** | http://www.conpaas.eu/dl/conpaas-2.1.0-rpi.img.tar.gz | MD5: 1eb2b8b369af3940460f85b12642f2e1 | size: 528 MB .. warning:: If you choose to use one of the images above, it is always a good idea to check its integrity before continuing to the next step. A corrupt image may result in unexpected behavior which may be hard to trace. You can check the integrity by verifying the MD5 hash with the ``md5sum`` command. Alternatively, you can also create one such image using the instructions provided in the section :ref:`image-creation`. The following steps will use as an example the image for the Raspberry PI. For other architecture or virtualization technologies, the commands are the same. .. warning:: The following steps need to be performed on a machine with the same architecture and a similar operating system. For the regular images, this means the 64-bit version of a Debian or Ubuntu system. For the Raspberry PI image, the steps need to be performed on the Raspberry PI itself (with a Raspbian installation, arm architecture). Trying to customize the Raspberry PI image on an x86 system will not work! #. Log in as **root** and change to the directory where you downloaded the image. #. Decompress the downloaded image:: root@raspberrypi:/home/pi# tar xaf conpaas-rpi.img.tar.gz #. (Optional) If you need to expand the size of the image, you can do it right now. As the image is in the raw format, expanding the size can be done by increasing the size of the image file. For example, to increase the size with 1 GB:: root@raspberrypi:/home/pi# dd if=/dev/zero bs=4M count=256 >> conpaas-rpi.img 256+0 records in 256+0 records out 1073741824 bytes (1.1 GB) copied, 56.05551 s, 19 MB/s If you have the package ``qemu-utils`` installed, you can also use ``qemu-img`` instead:: root@raspberrypi:/home/pi# qemu-img resize conpaas-rpi.img +1G Image resized. #. Map a loop device to the ConPaaS image:: root@raspberrypi:/home/pi# losetup -fv conpaas-rpi.img Loop device is /dev/loop0 .. warning:: If you already have other loop devices in use, the output of this command may contain a different loop device. Take a note of it and replace *loop0* with the correct device in the following commands. #. If you increased the size of the image in step 3, you now need to also expand the file system. First, check the integrity of the filesystem with the following command:: root@raspberrypi:/home/pi# e2fsck -f /dev/loop0 e2fsck 1.42.9 (4-Feb-2014) Pass 1: Checking inodes, blocks, and sizes Pass 2: Checking directory structure Pass 3: Checking directory connectivity Pass 4: Checking reference counts Pass 5: Checking group summary information root: 44283/117840 files (9.1% non-contiguous), 409442/470528 blocks You can now expand the file system:: root@raspberrypi:/home/pi# resize2fs /dev/loop0 resize2fs 1.42.9 (4-Feb-2014) Resizing the filesystem on /dev/loop0 to 732672 (4k) blocks. The filesystem on /dev/loop0 is now 732672 blocks long. #. Create a new directory and mount the image to it:: root@raspberrypi:/home/pi# mkdir conpaas-img root@raspberrypi:/home/pi# mount /dev/loop0 conpaas-img/ Now you can access the contents of the image inside the ``conpaas-img`` directory. #. Copy your application's binaries and any other static content that you want to include in the image somewhere under the ``conpaas-img`` directory. #. To install any prerequisites, you may want to change the root directory to ``conpaas-img``. But first, you will need to mount ``/dev``, ``/dev/pts`` and ``/proc`` in the ``conpaas-img`` directory (which will become the new root directory), or else the installation of some packages may fail:: root@raspberrypi:/home/pi# mount -obind /dev conpaas-img/dev root@raspberrypi:/home/pi# mount -obind /dev/pts conpaas-img/dev/pts root@raspberrypi:/home/pi# mount -t proc proc conpaas-img/proc #. You can now execute the chroot:: root@raspberrypi:/home/pi# chroot conpaas-img Your root directory is now the root of the image. #. To use *apt-get*, you need to set a working DNS server:: root@raspberrypi:/# echo "nameserver 8.8.8.8" > /etc/resolv.conf This example uses the Google Public DNS; you may, however, use any DNS server you prefer. Check that the Internet works in this new environment:: root@raspberrypi:/# ping www.conpaas.eu PING carambolier.irisa.fr (131.254.150.34) 56(84) bytes of data. 64 bytes from carambolier.irisa.fr (131.254.150.34): icmp_seq=1 ttl=50 time=35.8 ms [... output omitted ...] #. Use *apt-get* to install any packages that your application requires:: root@raspberrypi:/# apt-get update Hit http://archive.raspbian.org wheezy Release.gpg Hit http://archive.raspbian.org wheezy Release [... output omitted ...] root@raspberrypi:/# apt-get install <...> #. Make the final configurations (if needed) and make sure that everything works. #. Clean-up: Exit the chroot:: root@raspberrypi:/# exit exit root@raspberrypi:/home/pi# Unmount ``/dev``, ``/dev/pts`` and ``/proc``:: root@raspberrypi:/home/pi# umount conpaas-img/proc root@raspberrypi:/home/pi# umount conpaas-img/dev/pts root@raspberrypi:/home/pi# umount conpaas-img/dev Unmount the image:: root@raspberrypi:/home/pi# umount conpaas-img Remove the directory:: root@raspberrypi:/home/pi# rm -r conpaas-img Delete the loop device mapping:: root@raspberrypi:/home/pi# losetup -d /dev/loop0 That's it! Now the file ``conpaas-rpi.img`` contains the new ConPaaS image with your application pre-installed. You can now register the new image to the cloud of your choice and update the ConPaaS Director's settings to use the new image. Instructions are available in the Installation guide: **For Amazon EC2:** :ref:`registering-image-on-ec2` **For OpenStack:** :ref:`registering-image-on-openstack`