I am running my ruby on rails e-commerce application running on couple of ec2 instances in the production. I have not enabled auto scaling since we do Continuous Integration, we do deployment when ever required. I using chef for the all the reasons. I am trying to figure out how to use chef to deploy code to the multiple ec2 instances automcatically and manual intervention. I tried to deploy the code during the initial ec2 instance works fine. But my question is how do I do it in the auto scaling mode so that instance pull the latest code from either github or bit bucket and deployes itseld after bundle updating and all.
Add a comment
|
1 Answer
Fair enough. This is what i do and it works fine:
I have created an AMI with below settings:
install chef client
Create
/etc/chefdirectory- Copy
your_company-validator.pemfrom chef workstation to AMI at/etc/chef/validation.pem - Create
/etc/chef/client.rbas below.
If you use chef-solo, make the changes accordingly.
cat >> /etc/chef/client.rb <EF
log_level :auto
log_location STDOUT
chef_server_url "https://api.opscode.com/organizations/your_company"
validation_client_name "you_company-validator"
EF
Now, you have all above stuff configured in the AMI for chef bootstrap.
When you launch the AMI (using autoscaling or any other way), provide user-data where in it will run the chef-client with run_list of your choice. I provide below user-data:
#!/bin/bash
cat > /etc/chef/firstboot.json << EOL
{"run_list": ["recipe[java::oracle]"]}
EOL
chef-client -j /etc/chef/firstboot.json > /tmp/initialize_client.log 2>&1
basically, I am creating a .json file where I would specify the recipe/role/run_list I want to run. Once you provide the user-data, it will get executed on the 1st boot and chef-client will be run along with the run_list.
I think this is what you are looking for. so to summarize:
- Create an AMI with pre-installed
chef-client, client.rb and validation.pem - and then provide
user-dataat the time of instance launch.
This works perfectly for me. Let me know if this set-up gives you any trouble. I have been using this for quite some time.
So when I launch my instance using autscaling, it gets automatically chef-bootstrapped along with the recipes of my choice.
ASSUMPTION: you have all your recipes/roles created on chef-server before doing this. else the bootstrap will eventually fail.
7 Comments
Jeevan Dongre
Hello @slayedbylucifer sounds amazing!!! I will try it out soon and keep u updated
Jeevan Dongre
Hi I tried the process which you suggested but, after booting up ec2 instance it throws an error Failed to authenticate to the chef server (http 401). Server Response: ---------------- Failed to authenticate as 'ip-10-128-94-138.ap-southeast-1.compute.internal'. Ensure that your node_name and client key are correct. If these settings are correct, your client_key may be invalid. How I am suppose to debug it and what would be problem.
slayedbylucifer
when the chef-client runs for the 1st time, it would create client.pem itself. was there any client.pem in your image left from any prior run of chef-client? Also, Do you have your validation.pem key copied on the node ?
Mark O'Connor
I don't understand why create AMIs with chef pre-installed... The "knife bootstrap" command will install chef and automatically configure the /etc/chef directory. Better still you could use the ec2 plugin from knife and additionally spin up the VM.
slayedbylucifer
@Mark O'Connor,
knife bootstrap or knife ec2 plugin is no help when you are spinning up instances via AWS autoscalling. |