Let's build a Deis cluster in Google's Compute Engine!
Get a few Google things squared away so we can provison VM instances.
Install the Google Cloud SDK from https://developers.google.com/compute/docs/gcutil/#install. You will then need to login with your Google Account:
$ gcloud auth login
Your browser has been opened to visit:
https://accounts.google.com/o/oauth2/auth?redirect_uri=http%3A%2F%2Flocalhost%3A8085%2F&prompt=select_account&response_type=code&client_id=22535940678.apps.googleusercontent.com&scope=https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fappengine.admin+https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fbigquery+https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fcompute+https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdevstorage.full_control+https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fuserinfo.email+https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fndev.cloudman+https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fcloud-platform+https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fsqlservice.admin+https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fprediction+https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fprojecthosting&access_type=offline
You are now logged in as [youremail@gmail.com].
Your current project is [named-mason-824]. You can change this setting by running:
$ gcloud config set project <project>
Create a new project in the Google Developer Console (https://console.developers.google.com/project). You should get a project ID like orbital-gantry-285
back. We'll set it as the default for the SDK tools:
$ gcloud config set project orbital-gantry-285
Please note that you will begin to accrue charges once you create resources such as disks and instances
Navigate to the project console and then the Billing & Settings section in the browser. Click the Enable billing button and fill out the form. This is needed to create resources in Google's Compute Engine.
Google Computer Engine won't be available via the command line tools until it is initialized in the web console. Navigate to COMPUTE -> COMPUTE ENGINE -> VM Instances in the project console. The Compute Engine will take a moment to initialize and then be ready to create resources via gcutil
.
Create your cloud init file. It will look something like (be sure to generate and replace your own discovery URL):
#cloud-config
coreos:
etcd:
# generate a new token for each unique cluster from https://discovery.etcd.io/new
discovery: https://discovery.etcd.io/<token>
# multi-region and multi-cloud deployments need to use $public_ipv4
addr: $private_ipv4:4001
peer-addr: $private_ipv4:7001
units:
- name: etcd.service
command: start
- name: fleet.service
command: start
- name: format-ephemeral.service
command: start
content: |
[Unit]
Description=Formats the ephemeral drive
[Service]
Type=oneshot
RemainAfterExit=yes
ExecStart=/usr/sbin/wipefs -f /dev/disk/by-id/scsi-0Google_PersistentDisk_coredocker
ExecStart=/usr/sbin/mkfs.btrfs -f /dev/disk/by-id/scsi-0Google_PersistentDisk_coredocker
- name: var-lib-docker.mount
command: start
content: |
[Unit]
Description=Mount ephemeral to /var/lib/docker
Requires=format-ephemeral.service
Before=docker.service
[Mount]
What=/dev/disk/by-id/scsi-0Google_PersistentDisk_coredocker
Where=/var/lib/docker
Type=btrfs
- name: write-deis-motd.service
command: start
content: |
[Unit]
Description=write the Deis motd
ConditionFileNotEmpty=/run/deis/motd
[Service]
Type=oneshot
ExecStartPre=/usr/bin/rm /etc/motd
ExecStart=/usr/bin/ln -s /run/deis/motd /etc/motd
- name: link-deis-bashrc.service
command: start
content: |
[Unit]
Description=remove .bashrc file for CoreOS user
ConditionFileNotEmpty=/run/deis/.bashrc
[Service]
Type=oneshot
ExecStartPre=/usr/bin/rm /home/core/.bashrc
ExecStart=/usr/bin/ln -s /run/deis/.bashrc /home/core/.bashrc
write_files:
- path: /etc/deis-release
content: |
DEIS_RELEASE=v0.10.0
- path: /run/deis/motd
content: " \e[31m* * \e[34m* \e[32m***** \e[39mddddd eeeeeee iiiiiii ssss\n\e[31m* * \e[34m* * \e[32m* * \e[39md d e e i s s\n \e[31m* * \e[34m***** \e[32m***** \e[39md d e i s\n\e[32m***** \e[31m* * \e[34m* \e[39md d e i s\n\e[32m* * \e[31m* * \e[34m* * \e[39md d eee i sss\n\e[32m***** \e[31m* * \e[34m***** \e[39md d e i s\n \e[34m* \e[32m***** \e[31m* * \e[39md d e i s\n \e[34m* * \e[32m* * \e[31m* * \e[39md d e e i s s\n\e[34m***** \e[32m***** \e[31m* * \e[39mddddd eeeeeee iiiiiii ssss\n\n\e[39mWelcome to Deis\t\t\tPowered by Core\e[38;5;45mO\e[38;5;206mS\e[39m\n"
- path: /run/deis/.bashrc
owner: core
content: |
source /usr/share/skel/.bashrc
function nse() {
sudo nsenter --pid --uts --mount --ipc --net --target $(docker inspect --format="{{ .State.Pid }}" $1)
}
- path: /run/deis/bin/get_image
permissions: 0755
content: |
#!/bin/bash
# usage: get_image <component_path>
IMAGE=`etcdctl get $1/image 2>/dev/null`
# if no image was set in etcd, we use the default plus the release string
if [ $? -ne 0 ]; then
RELEASE=`etcdctl get /deis/release 2>/dev/null`
# if no release was set in etcd, use the default provisioned with the server
if [ $? -ne 0 ]; then
source /etc/deis-release
RELEASE=$DEIS_RELEASE
fi
IMAGE=$1:$RELEASE
fi
# remove leading slash
echo ${IMAGE#/}
Save this file as gce-cloud-init.yaml
. We will use it in the VM instance creation.
Create a SSH key that we will use for Deis host communication:
$ ssh-keygen -q -t rsa -f ~/.ssh/deis -N '' -C deis
Create some persistent disks to use for /var/lib/docker
. The default root partition of CoreOS is only around 4 GB and not enough for storing Docker images and instances. The following creates 3 disks sized at 32 GB:
$ gcutil adddisk --zone us-central1-a --size_gb 32 cored1 cored2 cored3
Table of resources:
+--------+---------------+--------+---------+
| name | zone | status | size-gb |
+--------+---------------+--------+---------+
| cored1 | us-central1-a | READY | 32 |
+--------+---------------+--------+---------+
| cored2 | us-central1-a | READY | 32 |
+--------+---------------+--------+---------+
| cored3 | us-central1-a | READY | 32 |
+--------+---------------+--------+---------+
Launch 3 instances using coreos-stable-367-1-0-v20140724
image. You can choose another starting CoreOS image from the listing output of gcloud compute images list
:
$ for num in 1 2 3; do gcutil addinstance --image coreos-stable-367-1-0-v20140724 --persistent_boot_disk --zone us-central1-a --machine_type n1-standard-2 --tags deis --metadata_from_file user-data:gce-cloud-config.yaml -disk cored${num},deviceName=coredocker --authorized_ssh_keys=core:~/.ssh/deis.pub,core:~/.ssh/google_compute_engine.pub core${num}; done
Table of resources:
+-------+---------------+--------------+---------------+---------+
| name | network-ip | external-ip | zone | status |
+-------+---------------+--------------+---------------+---------+
| core1 | 10.240.33.107 | 23.236.59.66 | us-central1-a | RUNNING |
+-------+---------------+--------------+---------------+---------+
| core2 | 10.240.94.33 | 108.59.80.17 | us-central1-a | RUNNING |
+-------+---------------+--------------+---------------+---------+
| core3 | 10.240.28.163 | 108.59.85.85 | us-central1-a | RUNNING |
+-------+---------------+--------------+---------------+---------+
We will need to load balance the Deis routers so we can get to Deis services (controller and builder) and our applications.
$ gcutil addhttphealthcheck basic-check --request_path /health-check
$ gcutil addtargetpool deis --health_checks basic-check --region us-central1 --instances core1,core2,core3
$ gcutil addforwardingrule deisapp --region us-central1 --target_pool deis
Table of resources:
+---------+-------------+--------------+
| name | region | ip |
+---------+-------------+--------------+
| deisapp | us-central1 | 23.251.153.6 |
+---------+-------------+--------------+
Note the forwarding rule external IP address. We will use it as the Deis login endpoint in a future step. Now allow the ports on the CoreOS nodes:
$ gcutil addfirewall deis-router --target_tags deis --allowed "tcp:80,tcp:2222"
We can create DNS records in Google Cloud DNS using the gcloud
utility. In our example we will be using the domain name rovid.io
. Create the zone:
$ gcloud dns managed-zone create --dns_name rovid.io. --description "Example Deis cluster domain name" rovidio
Creating {'dnsName': 'rovid.io.', 'name': 'rovidio', 'description':
'Example Deis cluster domain name'} in eco-theater-654
Do you want to continue (Y/n)? Y
{
"creationTime": "2014-07-28T00:01:45.835Z",
"description": "Example Deis cluster domain name",
"dnsName": "rovid.io.",
"id": "1374035518570040348",
"kind": "dns#managedZone",
"name": "rovidio",
"nameServers": [
"ns-cloud-d1.googledomains.com.",
"ns-cloud-d2.googledomains.com.",
"ns-cloud-d3.googledomains.com.",
"ns-cloud-d4.googledomains.com."
]
}
Note the nameServers
array from the JSON output. We will need to setup our upstream domain name servers to these.
Now edit the zone to add the Deis endpoint and wildcard DNS:
$ gcloud dns records --zone rovidio edit
{
"additions": [
{
"kind": "dns#resourceRecordSet",
"name": "rovid.io.",
"rrdatas": [
"ns-cloud-d1.googledomains.com. dns-admin.google.com. 2 21600 3600 1209600 300"
],
"ttl": 21600,
"type": "SOA"
}
],
"deletions": [
{
"kind": "dns#resourceRecordSet",
"name": "rovid.io.",
"rrdatas": [
"ns-cloud-d1.googledomains.com. dns-admin.google.com. 1 21600 3600 1209600 300"
],
"ttl": 21600,
"type": "SOA"
}
]
}
You will want to add two records as JSON objects. Here is an example edit for the two A record aditions:
{
"additions": [
{
"kind": "dns#resourceRecordSet",
"name": "rovid.io.",
"rrdatas": [
"ns-cloud-d1.googledomains.com. dns-admin.google.com. 2 21600 3600 1209600 300"
],
"ttl": 21600,
"type": "SOA"
},
{
"kind": "dns#resourceRecordSet",
"name": "deis.rovid.io.",
"rrdatas": [
"23.251.153.6"
],
"ttl": 21600,
"type": "A"
},
{
"kind": "dns#resourceRecordSet",
"name": "*.dev.rovid.io.",
"rrdatas": [
"23.251.153.6"
],
"ttl": 21600,
"type": "A"
}
],
"deletions": [
{
"kind": "dns#resourceRecordSet",
"name": "rovid.io.",
"rrdatas": [
"ns-cloud-d1.googledomains.com. dns-admin.google.com. 1 21600 3600 1209600 300"
],
"ttl": 21600,
"type": "SOA"
}
]
}
Time to install Deis!
Clone and checkout a version of Deis. In this example we will be deploying version 0.10.0
:
$ git clone https://github.com/deis/deis.git deis
remote: Counting objects: 3180, done.
remote: Compressing objects: 100% (1550/1550), done.
remote: Total 3180 (delta 1680), reused 2779 (delta 1419)
Receiving objects: 100% (3180/3180), 1.16 MiB | 565.00 KiB/s, done.
Resolving deltas: 100% (1680/1680), done.
From https://github.com/deis/deis
$ cd deis
$ git checkout v0.10.0
Note: checking out 'v0.10.0'.
You are in 'detached HEAD' state. You can look around, make experimental
changes and commit them, and you can discard any commits you make in this
state without impacting any branches by performing another checkout.
If you want to create a new branch to retain commits you create, you may
do so (now or later) by using -b with the checkout command again. Example:
git checkout -b new_branch_name
HEAD is now at 64708ab... chore(docs): update CLI versions and download links
Then install the CLI:
$ sudo pip install --upgrade ./client
The FLEETCTL_TUNNEL
environment variable provides a SSH gateway to use in with CoreOS. Use the public IP address for one of the CoreOS nodes we deployed earlier:
export FLEETCTL_TUNNEL=23.236.59.66
Verify the CoreOS cluster is operation and that we can see all our nodes:
$ fleetctl list-machines
MACHINE IP METADATA
b0d96509... 10.240.19.74 -
c6e39062... 10.240.174.34 -
e084a014... 10.240.26.179 -
Now we can bootstrap the Deis containers. DEIS_NUM_INSTANCES
should match the number of EC2 instances launched. DEIS_NUM_ROUTERS
should be at least 2. But can also match node count:
DEIS_NUM_INSTANCES=3 DEIS_NUM_ROUTERS=3 make run
Job deis-router.1.service loaded on 47c540a2.../10.240.62.89
Job deis-router.2.service loaded on 794c2897.../10.240.194.149
Job deis-router.3.service loaded on 177b5a76.../10.240.98.27
Job deis-builder-data.service loaded on 177b5a76.../10.240.98.27
Job deis-database-data.service loaded on 47c540a2.../10.240.62.89
Job deis-logger-data.service loaded on 177b5a76.../10.240.98.27
Job deis-registry-data.service loaded on 47c540a2.../10.240.62.89
fleetctl --strict-host-key-checking=false load registry/systemd/deis-registry.service logger/systemd/deis-logger.service cache/systemd/deis-cache.service database/systemd/deis-database.service
Job deis-logger.service loaded on 177b5a76.../10.240.98.27
Job deis-registry.service loaded on 47c540a2.../10.240.62.89
Job deis-cache.service loaded on 47c540a2.../10.240.62.89
Job deis-database.service loaded on 47c540a2.../10.240.62.89
fleetctl --strict-host-key-checking=false load controller/systemd/*.service
Job deis-controller.service loaded on 177b5a76.../10.240.98.27
fleetctl --strict-host-key-checking=false load builder/systemd/*.service
Job deis-builder.service loaded on 177b5a76.../10.240.98.27
Deis components may take a long time to start the first time they are initialized.
Waiting for 1 of 3 deis-routers to start...
fleetctl --strict-host-key-checking=false start -no-block deis-router.1.service; fleetctl --strict-host-key-checking=false start -no-block deis-router.2.service; fleetctl --strict-host-key-checking=false start -no-block deis-router.3.service;
Triggered job deis-router.1.service start
Triggered job deis-router.2.service start
Triggered job deis-router.3.service start
Waiting for deis-registry to start...
fleetctl --strict-host-key-checking=false start -no-block registry/systemd/deis-registry.service logger/systemd/deis-logger.service cache/systemd/deis-cache.service database/systemd/deis-database.service
Triggered job deis-registry.service start
Triggered job deis-logger.service start
Triggered job deis-cache.service start
Triggered job deis-database.service start
Waiting for deis-controller to start...
fleetctl --strict-host-key-checking=false start -no-block controller/systemd/*
Triggered job deis-controller.service start
Waiting for deis-builder to start...
fleetctl --strict-host-key-checking=false start -no-block builder/systemd/*.service
Triggered job deis-builder.service start
Your Deis cluster is ready to go! Continue following the README to login and use Deis.
This operation will take a while as all the Deis systemd units are loaded into the CoreOS cluster and the Docker images are pulled down. Grab some iced tea!
Verify that all the units are active after the make run
operation completes:
$ fleetctl list-units
UNIT STATE LOAD ACTIVE SUB DESC MACHINE
deis-builder-data.service loaded loaded active exited deis-builder-data 177b5a76.../10.240.98.27
deis-builder.service launched loaded active running deis-builder 177b5a76.../10.240.98.27
deis-cache.service launched loaded active running deis-cache 47c540a2.../10.240.62.89
deis-controller.service launched loaded active running deis-controller 177b5a76.../10.240.98.27
deis-database-data.service loaded loaded active exited deis-database-data 47c540a2.../10.240.62.89
deis-database.service launched loaded active running deis-database 47c540a2.../10.240.62.89
deis-logger-data.service loaded loaded active exited deis-logger-data 177b5a76.../10.240.98.27
deis-logger.service launched loaded active running deis-logger 177b5a76.../10.240.98.27
deis-registry-data.service loaded loaded active exited deis-registry-data 47c540a2.../10.240.62.89
deis-registry.service launched loaded active running deis-registry 47c540a2.../10.240.62.89
deis-router.1.service launched loaded active running deis-router 47c540a2.../10.240.62.89
deis-router.2.service launched loaded active running deis-router 794c2897.../10.240.194.149
deis-router.3.service launched loaded active running deis-router 177b5a76.../10.240.98.27
Everything looks good! Register the admin user. The first user added to the system becomes the admin:
$ deis register http://deis.rovid.io
username: andyshinn
password:
password (confirm):
email: andys@andyshinn.as
Registered andyshinn
Logged in as andyshinn
You are now registered and logged in. Create a new cluster named dev
to run applications under. You could name this cluster something other than dev
. We only use it as an example to illustrate a cluster can be restricted to certain CoreOS nodes. The hosts are the internal IP addresses of the CoreOS nodes:
$ deis clusters:create dev rovid.io --hosts 10.240.33.107,10.240.94.33,10.240.28.163 --auth ~/.ssh/deis
Creating cluster... done, created dev
Add your SSH key so you can publish applications:
$ deis keys:add
Found the following SSH public keys:
1) id_rsa.pub andy
Which would you like to use with Deis? 1
Uploading andy to Deis...done
Creating an application requires that application be housed under git already. Clone an example application and deploy it:
$ git clone https://github.com/deis/example-ruby-sinatra.git
Cloning into 'example-ruby-sinatra'...
remote: Counting objects: 98, done.
remote: Compressing objects: 100% (50/50), done.
remote: Total 98 (delta 42), reused 97 (delta 42)
Unpacking objects: 100% (98/98), done.
Checking connectivity... done.
$ cd example-ruby-sinatra
$ deis
$ deis create
Creating application... done, created breezy-frosting
Git remote deis added
Time to push:
$ git push deis master
Counting objects: 98, done.
Delta compression using up to 8 threads.
Compressing objects: 100% (92/92), done.
Writing objects: 100% (98/98), 20.95 KiB | 0 bytes/s, done.
Total 98 (delta 42), reused 0 (delta 0)
-----> Ruby app detected
-----> Compiling Ruby/Rack
-----> Using Ruby version: ruby-1.9.3
-----> Installing dependencies using 1.6.3
Running: bundle install --without development:test --path vendor/bundle --binstubs vendor/bundle/bin -j4 --deployment
Don't run Bundler as root. Bundler can ask for sudo if it is needed, and
installing your bundle as root will break this application for all non-root
users on this machine.
Fetching gem metadata from http://rubygems.org/..........
Fetching additional metadata from http://rubygems.org/..
Using bundler 1.6.3
Installing rack 1.5.2
Installing tilt 1.3.6
Installing rack-protection 1.5.0
Installing sinatra 1.4.2
Your bundle is complete!
Gems in the groups development and test were not installed.
It was installed into ./vendor/bundle
Bundle completed (5.72s)
Cleaning up the bundler cache.
-----> Discovering process types
Procfile declares types -> web
Default process types for Ruby -> rake, console, web
-----> Compiled slug size is 12M
remote: -----> Building Docker image
remote: Sending build context to Docker daemon 11.77 MB
remote: Sending build context to Docker daemon
remote: Step 0 : FROM deis/slugrunner
remote: ---> f607bc8783a5
remote: Step 1 : RUN mkdir -p /app
remote: ---> Running in dd1cb10534c0
remote: ---> 3151b07f7623
remote: Removing intermediate container dd1cb10534c0
remote: Step 2 : ADD slug.tgz /app
remote: ---> b86143c577ae
remote: Removing intermediate container 63dca22b29d6
remote: Step 3 : ENTRYPOINT ["/runner/init"]
remote: ---> Running in 43c572eacc69
remote: ---> 6eeace9fea7e
remote: Removing intermediate container 43c572eacc69
remote: Successfully built 6eeace9fea7e
remote: -----> Pushing image to private registry
remote:
remote: Launching... done, v2
remote:
remote: -----> breezy-frosting deployed to Deis
remote: http://breezy-frosting.dev.rovid.io
remote:
remote: To learn more, use `deis help` or visit http://deis.io
remote:
To ssh://git@deis.rovid.io:2222/breezy-frosting.git
* [new branch] master -> master
Your application will now be built and run inside the Deis cluster! After the application is pushed it should be running at http://breezy-frosting.dev.rovid.io. Check the application information:
$ deis apps:info
=== breezy-frosting Application
{
"updated": "2014-07-28T00:35:45.528Z",
"uuid": "fd926c94-5b65-48e8-8afe-7ac547c12bd6",
"created": "2014-07-28T00:33:35.346Z",
"cluster": "dev",
"owner": "andyshinn",
"id": "breezy-frosting",
"structure": "{\"web\": 1}"
}
=== breezy-frosting Processes
--- web:
web.1 up (v2)
=== breezy-frosting Domains
No domains
Can we connect to the application?
$ curl -s http://breezy-frosting.dev.rovid.io
Powered by Deis!
It works! Enjoy your Deis cluster in Google Compute Engine!
Cheers, I'll have a look at it tomorrow GMT+7. I'll let you know how it goes. Cheers