Over the last few months I've become familiar with the AWS OpsWorks deployment process as it pertain to Node.js - deployment for Go seems to be another animal.
From what I've gathered, this is what I need to compile a successful Go deployment:
Install go on the EC2 box
Pull the private repository from GitHub
Pull in all dependencies
Compile the main package for the box's arch
Start the binary with a couple of flags that I use
Everywhere I have read seems to tout the ease of Go deployments because dependencies are included in the binary, but that seems to imply that you are compiling the application in your development environment and pushing that up to the cloud. This doesn't seem like a process that works well across a development team.
https://github.com/crowdmob/chef-golang-web-server-cookbook
I have been attempting to get the Chef Scripts from CrowdMob working, but to no avail. I continue to get errors that look like this:
[2014-08-01T16:08:22+00:00] WARN: Cookbook 'templates' is empty or entirely chefignored at /opt/aws/opsworks/current/merged-cookbooks/templates
What is the proper way to deal with dependencies during deployment?
Are there any established practices for deploying Go onto AWS with Chef?
Use a continuous integration service like CircleCi, Travis or your own setup Jenkins.
On the Continuous integration service then
Add a github post commit hook .
Test / Build the binary
Create the zip file as artifact
At this point you can create an new version on Elastic Beanstalk using the AWS commandline and the zip file created from this version.
venv/bin/aws elasticbeanstalk create-application-version ...
Then just select which version to deploy from the EB dashboard.
For simple services using Chef is overkill IMHO. Docker offers a simple workflow.
Use the Docker container option and then use elastic beanstalk's command-line client to initialize your environment in the project root directory and then you can simply do a 'git aws.push' from the same place.
With the correctly configured Dockerfile in your project and pushed to eb, the EBS' docker container app will pull the correct image with golang installed, then do a go get on your projects dependencies, and then compile and run your app. It sounds way more complicated but it's actually very easy.
Below is a link to a video walkthrough I did for running a simple golang webapp on EBS. The method for uploading the project does not use git. Instead, I zip it up and upload it, but the git method is recommended (and I do it) for automating deployment.
YouTube: How to run a go web app on Amazon's Elastic Beanstalk
I also had some problems to setup a good building process with Elastic Beanstalk and Go.. I don't want to use Docker, and all the people seems to be going on this direction.. so.. you can take a look at this project: https://github.com/battle-arena/heimdall
There you will find a custom setup using the Buildfile and the Procfile.. and I rely on a CI system to build the release package...
Basically I do the following:
Hook the commits to a CI system
On the CI system I run the test and the install.sh if all good
The install.sh will create a build folder and a structure that will be sended to the Elastic Beanstalk with the aws-cli tool
After send to the EB the Buildfile will run the build.sh that will basically extract the compressed package with the proper structure, and run a go get ./... and go build
The Procfile will run the generated binary
I think the result is pretty good, and you can use with any CI tool.
Related
I tried to launch our Scala application on the Swisscom Cloud Foundry (CF) infrastructure. To do so, the matching Heroku buildpack was used:
https://github.com/heroku/heroku-buildpack-scala
As this did not work, I tried to deploy the 'hello-scala' example using this buildpack.
My fork to be able to build the slightly outdated example:
https://github.com/AlwinEgger/hello-scala
I have to underline that I am fetching the port I have to use as env variable 'PORT'.
Unfortunately, there is not much on the log. "failed to accept connections within health check timeout" message indicates that there is no one listening...
My questions: Did anyone succeed in deploying Scala apps on CF infrastructures (# Swisscom)?
A workaround I found:
I'm not using the scala- but the java-buildback. This with the major advantage and inconvenience that the project is not any more build on the instance.
Advantage: It speeds up the whole process considerably
Inconvenience: A build server is needed
So what do we have to do?
An example may be found here (this is the actual application):
https://github.com/OpenOlitor/openolitor-server
Add the sbt-native-packager to your project
Execute the action 'universal:packageBin' building by hand or configure your build server to do so
Change the buildpack in the manifest.yml and add some parameters, if necessary. Configure the path of the artifact to deploy.
Run cf push or let the build server do so.
We are currently using web deploy for creating packages for our .net web applications. It got some pros and cons. Now we are going to use nuget for dependency management but given the ease of packaging in nuget .. i am debating on should i still use web deploy + remote service or try to use nuget to bundle my web application and use powershell or something like octupus to deploy ??
For my web deploy becomes little complex when even try to do simple things like include , exlcude , gac or registry or iis config ( which again not very flexible ).
But on the other hand it comes with remote service and all i need to do is through the package to the service and i am done..
Please give your inputs on this comparison.
-raj
NuGet is a dependency manager, and as such is appropriate for use at development/build, not deployment time. From a deployment perspective, it doesn't offer you any more than what a zip file does... except all the overhead of trying to fit NuGet in the process.
Take a look at Inedo's BuildMaster. It can take care of the process from source control through production deployment. There's also a free version that will most certainly handle your requirements, and it also has a module to manage your configuration files so you don't need to worry about doing transforms.
(disclaimer: I work for Inedo)
I am setting up a system that uses NuGet and Webdeploy.
NuGet is used as the repository format, so the build system publishes NuGet packaged artifacts.
The Deployment systemn uses NuGet to get the RIGHT packaged artifact from the NuGet repository.
The build artififact happens to be WebDeploy for easy installation.
I am still working out if I am going to use puppet, chef or octupus to orchastrate the deployment.
I'm using Fedora and I deploy Symfony projects in my local machine using virtual hosts. How can I deploy my projects in server to public which others can view it through their machines?
Thanx...
You have several way to deploy you symfony project. I will avoid ftp, svn up on prod, etc .. So, here is 2 good ways.
The built-in deploy task
Symfony comes with a built-in depoy task that has been used when symfony 1.4 was released. I think it's less and less used now (because there is better tool).
The simplest way to deploy your website is to use the built-in project:deploy task. It uses SSH and rsync to connect and transfer the files from one computer to another one.
Using capifony, which use Capistrano
Capistrano is an open source tool for running scripts on multiple servers. It’s primary use is for easily deploying applications.
capifony is a deployment recipes collection that works with both symfony and Symfony2 applications.
This way is far better than the previous one because you can automate many script when deploying (like testing your code, start a fresh built lib, upgrade database, share config file). But the most important one (from my POV) is that you can easily rollback a bad deployement. It's damm easy.
How do you handle the deployment of a LightSwitch application into a production environment?
i.e. the LS application has been developed, but it now needs to be installed first into Test, and then into Live.
We don't want to use the "manual" approach, i.e. use the Visual Studio Build / Publish option, rather we want to automate the deployment.
My feeling is that deployment is one of the real weak points of LightSwitch. If you are using the very simple deployment model that is build into the product, and you're doing everything within a Windows domain, the publishing wizard can do everything. But if you're deviating from the model at all LightSwitch will fight you. I'd really like to see an "advanced" deployment option that provided some configurability.
Here's how I solved the problem you're having with LightSwitch applications that are targeting web deployment:
At the beginning of the project, deploy once to each target environment using the publish wizard. This is the easiest way to get the database set up.
As new builds are deployed, use the publish wizard to deploy to a deployment package to a standard location on the local development machine.
The deployment package is just a zip file, so you can open it an drill down to where the actual binary release is. I use a powershell script to copy the binary files out of the the deployment package and in to a local SVN working directory. Note that you must not copy web.config file during this step.
Check the unpacked binary files into SVN and use SVN to manage the deployment.
Manage schema changes with SQL scripts.
How do you deploy your websites?
For example: I am developing a site with a php framework and have it under version controle with git with all my local configs. When I want to put it on a web server for testing or updating the live application i have to copy it onto the server, change the config files, delete my test stuff etc.
So how do you handle these tasks?
I thought about using ant and write a deployment script for this.
Does there already exists a common solution for this "problem"? Because I don't think im the only one who need something like this.
There are quite a bunch of stuff available, but you might like Phing (like ANT for java).
Questions related to PHP+Phing:
Do you use Phing?
How do you manage your build [using Phing] process?
Setting up a deployment / build / CI cycle for PHP projects
what can Phing do that Ant can't?
Also read this questions sounds very interesting How To Deploy Your PHP Applications Correctly?
There is a specific question (a possible duplicate of your questions) has been answered a while ago, take a look at it Deploy a project using Git push
It seems you are using php, you should be good to go with capistrano. It is very easy to use capistrano for deployment with rails but it can also be tricked a bit to use for php.
Basically what you do with capistrano is -
Tell it which is you application server
Tell you database server
Tell web server (in most cases web server, app server and db server are same)
Specify you git repository with branch you want to deploy from
Once configured, you can deploy with capistrano with single command. You can even rollback your deployments from some of backup releases created by capistrano.
Now form some the repetitive tasks like, copying configs files like database configs (which generally are ignored in git), you create some tasks, which just creates symlinks or copies the files at appropriate location. These tasks will be called with deploy_hookes e.g. after_symlink hook.
You can find more about capistrano here - https://github.com/capistrano/capistrano/wiki
It comes with very good documentation, after getting overview, you may search for your framework specific approach to do this.