Install Chef-Server 11 on EC2 Instance

Question

I am using hosted Chef for quite some time. Wanted to explore the opensource chef server. hence I am trying to setup my Chef-Server 11 on EC2 instance.

I have Chef-server running and I can access the web GUI for the same. I have the chef-workstation configured on another ec2 instance that is also working fine.

Problem: I am not able to upload any cookbook. I get below error when I try uploading the cookbook:

# knife cookbook upload getting-started
Uploading getting-started [0.4.0]
/opt/chef/embedded/lib/ruby/1.9.1/net/http.rb:763:in `initialize': Connection refused - connect(2) (Errno::ECONNREFUSED)

However, other list commands of knife are working fine.

I did my home work and bumped on below links:

http://www.opscode.com/blog/2013/03/11/chef-11-server-up-and-running/

http://www.curvve.com/blog/servers/2013/script-to-configure-and-set-your-hostname-and-fqdn-on-ec2-instances/

So,

It is mentioned that the chef-server needs a working FQDN to work. I set the my public ec2 host name as the hostname of the server as well as set it up in /etc/hosts. Rebooted the instance. Ran chef-server-ctl reconfigure again. And still facing the same error.

QUESTION: How to figure out the FQDN part of the EC2 instance for chef-server to work? if anyone has set up chef-server successfully on EC2 and was able to upload the cookbooks, then please share your steps for FQDN workout.

Answer 1

This is how i got it working. updated the public DNS name of my ec2 instance (chef-server) in /etc/sysconfig/network and service network restart. Now I am able to upload the cookbooks fine.

Need to think about elastic IP as potential option for my chef-server.

Answer 2

I was having a hard time with this but this solution worked!

Edit /etc/chef-server/chef-server.rb and add these lines (create the file if it doesn't exist):

server_name = "THE PUBLIC IP OF YOUR INSTANCE"
api_fqdn server_name
nginx['url'] = "https://#{server_name}"
nginx['server_name'] = server_name
lb['fqdn'] = server_name
bookshelf['vip'] = server_name

I found the solution here http://sahebjade.blogspot.com/2013/05/check-your-knife-configuration-and.html

Answer 3

Edit /etc/chef-server/chef-server.rb and add these lines (create the file if it doesn't exist):

bookshelf["vip"] = node["ipaddress"]
bookshelf["url"] = "https://#{node["ipaddress"]}"
erchef['s3_url_ttl'] = 3600

The first two lines will point your chef-server URL to the machine's IP and the third will solve a timeout issue that apparently always exist when the Chef Server is on EC2.

Answer 4

I wanted to expand some on the answers since they don't give a complete picture. This applies to Chef 11 (hopefully Chef 12 is smarter)

In my case I rolled a master up under VPC #1 which gave it an internal address like this

ip-10-0-0-10.ec2.internal

Because I was only playing with the VPC initially, I had misconfigured some things I needed so I had to drop it and I created a new scheme. Thankfully, I was able to snapshot the old Chef master and bring it up under the new VPC but I found that I couldn't log into Chef anymore. It took some digging but I found in my /var/log/chef-server/chef-server-webui/current log that the install had glommed onto the old hostname and set that as the internal URL for... everything. This caused problems after the internal hostname change

2014-12-24_16:19:09.46680 SocketError: Error connecting to https://ip-10-0-0-10.ec2.internal/users/admin - getaddrinfo: Name or service not known

Now, to the OP answer

Need to think about elastic IP as potential option for my chef-server

In my case, I just added a CNAME to CloudFlare and set that as my permanent address. Since I can set CloudFlare to a low TTL on that one address it makes it easy to move it around between IP changes (I don't need an Elastic IP while I'm just getting it configured). This way I could then tell Chef to always look for the same URL and not worry about an EIP.

Once that was done, I had to update Chef. I don't know what changed (this is 11.16.4) but I found the configs live in /var/opt/chef-server/chef-server-webui/etc/chefserver.rb as opposed to some of the other answers listing chef-server.rb. Not sure if that's a YMMV thing or not. I changed the following towards the bottom of that file

# Environment specific application configuration.
# These values override the ones set in 'RAILS_ROOT/config/application.rb'
#config.chef_server_url = "https://ip-10-0-0-10.ec2.internal"
config.chef_server_url = "https://chef.mydomain.com"

I also changed /var/opt/chef-server/nginx/etc/chef_https_lb.conf

server_name chef.mydomain.com;

Finally I restarted Chef

chef-server-ctl restart

That seems to have done the trick. Logins work again.