bosh-lite

A local development environment for BOSH using warden containers in a vagrant box.

This readme walks through deploying Cloud Foundry with bosh-lite. Bosh and bosh-lite can be used to deploy just about anything once you've got the hang of it.

Installation

For all use cases, first prepare this project with bundler .

1. Install vagrant

   Known working version:

   $ vagrant -v
   Vagrant 1.4.3
   

   See this blog for special instructures for Windows users of bosh-lite.

2. Install Ruby + RubyGems + Bundler

3. Run Bundler from the base directory of this repository

   bundle
   

Below are installation insructions for different Vagrant providers.

• VMWare Fusion
• Virtualbox
• AWS

Using the VMWare Fusion Provider (preferred)

Fusion is faster, more reliable and we test against it more frequently. Both fusion and the vagrant fusion provider require a license.

Known to work with Fusion version 6.0.2 and vagrant plugin vagrant-vmware-fusion version 2.2.0 .

1. Install vagrant Fusion Plugin and license

   This requires a license file for Fusion. If you don't have one visit http://www.vagrantup.com to purchase a license.

   vagrant plugin install vagrant-vmware-fusion
   vagrant plugin license vagrant-vmware-fusion license.lic
   

2. Start vagrant from the base directory of this repository (which uses the Vagrantfile)

   vagrant up --provider vmware_fusion
   

3. Bosh target (login with admin/admin)

   $ bosh target 192.168.50.4
   Target set to `Bosh Lite Director'
   $ bosh login
   Your username: admin
   Enter password: *****
   Logged in as `admin'
   

4. Add a set of route entries to your local route table to enable direct warden container access. Your sudo password may be required.

   scripts/add-route
   

Using the Virtualbox Provider

1. Start vagrant from the base directory of this repository (which uses the Vagrantfile)

   vagrant up
   

2. Bosh target (login with admin/admin)

   $ bosh target 192.168.50.4
   Target set to `Bosh Lite Director'
   $ bosh login
   Your username: admin
   Enter password: *****
   Logged in as `admin'
   

3. Add a set of route entries to your local route table to enable direct warden container access every time your networking gets reset (eg. reboot or connect to a different network). Your sudo password may be required.

   scripts/add-route
   

Using the AWS Provider

1. Install Vagrant AWS provider

   vagrant plugin install vagrant-aws
   

   Known to work for version: vagrant-aws 0.4.1

2. Add dummy AWS box

   vagrant box add dummy https://github.com/mitchellh/vagrant-aws/raw/master/dummy.box
   

3. Set environment variables called BOSH_AWS_ACCESS_KEY_ID and BOSH_AWS_SECRET_ACCESS_KEY with the appropriate values. If you've followed along with other documentation such as these steps to deploy Cloud Foundry on AWS, you may simply need to source your bosh_environment file.

4. Make sure the EC2 security group you are using in the Vagrantfile exists and allows inbound TCP traffic on ports 25555 (for the BOSH director), 22 (for SSH), 80/443 (for Cloud Controller), and 4443 (for Loggregator).

5. Run Vagrant from the aws folder:

   cd aws
   vagrant up --provider=aws
   cd ..
   

6. find out the public IP of the box you just launched. You can see this info at the end of vagrant up output. Another way is running vagrant ssh-config under aws folder.

7. Bosh target (login with admin/admin)

   $ bosh target <public_ip_of_the_box>
   Target set to `Bosh Lite Director'
   $ bosh login
   Your username: admin
   Enter password: *****
   Logged in as `admin'
   

8. Edit manifests/cf-stub-spiff.yml to include a 'domain' key under 'properties' that corresponds to a domain you've set up for this Cloud Foundry instance, or if you want to use xip.io, it can be {your.public.ip}.xip.io.

9. Direct future traffic received on the instance to another ip (the HAProxy):

sudo iptables -t nat -A PREROUTING -p tcp -d <internal IP of instance> --dport 80 -j DNAT --to 10.244.0.34:80
sudo iptables -t nat -A PREROUTING -p tcp -d <internal IP of instance> --dport 443 -j DNAT --to 10.244.0.34:443
sudo iptables -t nat -A PREROUTING -p tcp -d <internal IP of instance> --dport 4443 -j DNAT --to 10.244.0.34:4443

These rules are cleared on restart. They can be saved and configured to be reloaded on startup if so desired (granted the internal ip remains the same).

Restart the director

Occasionally you need to restart the bosh-lite director to avoid cloudfoundry-attic#82 ...) so perhaps always run the following after booting up bosh-lite:

vagrant ssh -c "sudo sv restart director"

Upload Warden stemcell

bosh-lite uses the Warden CPI, so we need to use the Warden Stemcell which will be the root file system for all Linux Containers created by the Warden CPI.

1. Download latest warden stemcell

   wget http://bosh-jenkins-gems-warden.s3.amazonaws.com/stemcells/latest-bosh-stemcell-warden.tgz
   

2. Upload the stemcell

   bosh upload stemcell latest-bosh-stemcell-warden.tgz
   

NB: It is possible to do this in one command instead of two, but doing this in two steps avoids having to download the stemcell again when you bring up a new bosh-lite box.

Note: You can also use 'bosh public stemcells' to list and download the latest warden stemcell

example (the versions you see will be different from these):

$ bosh public stemcells 
+---------------------------------------------+
| Name                                        |
+---------------------------------------------+
| bosh-stemcell-1722-aws-xen-ubuntu.tgz       |
| bosh-stemcell-1722-aws-xen-centos.tgz       |
| light-bosh-stemcell-1722-aws-xen-ubuntu.tgz |
| light-bosh-stemcell-1722-aws-xen-centos.tgz |
| bosh-stemcell-1722-openstack-kvm-ubuntu.tgz |
| bosh-stemcell-1722-vsphere-esxi-ubuntu.tgz  |
| bosh-stemcell-1722-vsphere-esxi-centos.tgz  |
| bosh-stemcell-24-warden-boshlite-ubuntu.tgz |
+---------------------------------------------+

$ bosh download public stemcell bosh-stemcell-24-warden-boshlite-ubuntu.tgz

Deploy Cloud Foundry

1. Install spiff. Use the latest 0.3 binary of spiff extract it and make sure that spiff is in your $PATH.

2. clone a copy of cf-release:

   cd ~/workspace
   git clone https://github.com/cloudfoundry/cf-release
   

3. Decide which final release of Cloud Foundry you wish to deploy, by looking at in the releases directory of cf-release. At the time of this writing, cf-149 is the most recent. We will use that as the example, but you are free to substitute any future release.

4. Check out the desired revision of cf-release, (eg, 149)

   cd ~/workspace/cf-release
   ./update
   git checkout v149
   

5. Use the make_manifest_spiff script to create a cf manifest. This step assumes you have cf-release checked out to ~/workspace [note that you can have it checked out to somewhere else, you just have to set the BOSH_RELEASES_DIR environment variable to something other than its default value of ~/workspace]. It requires that cf-release is checked out the tag matching the final release you wish to deploy so tha the templates used by make_manifest_spiff match the code you are deploying.

   make_manifest_spiff will target your bosh-lite director, find the uuid, create a manifest stub and run spiff to generate a manifest at manifests/cf-manifest.yml. (If this fails, try updating spiff)

   cd ~/workspace/bosh-lite
   ./scripts/make_manifest_spiff
   

6. Upload final release

   cd ~/workspace/cf-release
   bosh upload release releases/cf-149.yml
   

7. Deploy CF to bosh-lite

   cd ~/workspace/bosh-lite
   bosh deployment manifests/cf-manifest.yml
   bosh deploy
   # enter yes to confirm
   

8. Run the yeti tests against your new deployment to make sure it's working correctly.

   a. Set the environment variables VCAP_BVT_API_ENDPOINT, VCAP_BVT_ADMIN_USER, VCAP_BVT_ADMIN_USER_PASSWD

   Might look like this:

   # This is the HA Proxy ip in bosh vms (not the cc)
   export VCAP_BVT_API_ENDPOINT=http://api.10.244.0.34.xip.io
   export VCAP_BVT_ADMIN_USER=admin
   export VCAP_BVT_ADMIN_USER_PASSWD=admin
   

   If you do not want to use xip.io or are going to be offline, you can try the custom DNS and offline instructions for OSX.

   b. Run yeti as normal from cf-release/src/tests.. e.g. (Make sure you are logged in via cf before running these commands)

   bundle
   rake config:clear_bvt # clear the BVT from previous runs
   bundle exec rake prepare; # create initial users/assets
   bundle exec rspec # run!
   
   ./warden_rspec # Run tests in parallel
   

SSH into deployment jobs

To use bosh ssh to SSH into running jobs of a deployment, to run the following command:

scripts/add-route

Now you can now SSH into any VM with bosh ssh:

$ bosh ssh
1. nats/0
2. syslog_aggregator/0
3. postgres/0
4. uaa/0
5. login/0
6. cloud_controller/0
7. loggregator/0
8. loggregator-router/0
9. health_manager/0
10. dea_next/0
11. router/0
Choose an instance:

Restore your deployment

The warden container will be lost after vm reboot, but you can restore your deployment with bosh cck, bosh's command for recovering from unepexted errors.

$ bosh cck

Choose 2 to recreate each missing vm:

Problem 1 of 13: VM with cloud ID `vm-74d58924-7710-4094-86f2-2f38ff47bb9a' missing.
  1. Ignore problem
  2. Recreate VM using last known apply spec
  3. Delete VM reference (DANGEROUS!)
Please choose a resolution [1 - 3]: 2
...

Typing yes to confirm at the end:

Apply resolutions? (type 'yes' to continue): yes

Applying problem resolutions
  missing_vm 212: Recreate VM using last known apply spec (00:00:13)
  missing_vm 215: Recreate VM using last known apply spec (00:00:08)
...
Done                    13/13 00:03:48

Troubleshooting

1. Starting over again is often the quickest path to success, you can use vagrant destroy from the base directory of this project to remove the VM.
2. To start with a new VM just execute the appropriate vagrant up command optionally with the provider option as shown in the earlier sections.

Manage your local boxes

We publish pre-built Vagrant boxes on Amazon S3. It is recommened to use the latest boxes.

Download latest boxes

Just get a latest copy of the Vagrantfile from this repo and run vagrant up.

Delete old boxes

Free some disk space by deleting the old boxes.

$ vagrant box list
boshlite-ubuntu1204-build55 (virtualbox)
boshlite-ubuntu1204-build55 (vmware_desktop)
boshlite-ubuntu1204-build74 (virtualbox)
boshlite-ubuntu1204-build83 (virtualbox)

$ vagrant box remove boshlite-ubuntu1204-build55 virtualbox
Removing box 'boshlite-ubuntu1204-build55' with provider 'virtualbox'...