1. DevOps is hard
It might not seem like it, but DevOps is hard. A few years ago I thought to myself that it can’t be that difficult since installing an individual application isn’t that difficult. I was wrong in part because…
2. Security is hard
Production is scary. I’d rather not have access when possible. On the other hand the tools that we use will definitely need access to production since it’s kind of the reason they exist. This means we have to have very tight control over who has access to the credentials that the tools run under. We work to limit our own day-to-day accounts so that their access is limited as well.
As a developer I didn’t think much about Security. I pretty much just stuffed an AD Group in a config file somewhere when I was told to and I was done. As a DevOps engineer I have had (and will continue) to learn a lot more about security and its organization even though I don’t manage security for my organization. Security impacts deployments at every level so you will have to learn about security infrastructure in order to make safe and practical recommendations to your security administration group.
3. You are not Netflix (unless you are)
Our organization got excited about DevOps tools after seeing some compelling presentations by Netflix at QCon San Francisco. Netflix has the need for highly scaled web servers which fully embrace the “cattle vs. pets” philosophy because they have millions of concurrent users of a publicly facing service.
We are not Netflix. We have 50+ internal applications with usage rates measured in the 10’s. They’re important to us–they run our business–but our problems are not the same ones Netflix faces. The tools that Netflix uses are designed to solve problems Netflix has. That doesn’t necessarily make them a good fit for our needs. We lost a lot of time and effort trying to make Netflix solutions fit our problems.
4. Windows vs. Linux matters when choosing your tools.
There are basically 5 possibilities when it comes to your server topology:
- Windows Only
- Linux Only
- Windows Dominant
- Linux Dominant
- Hetergeneous
If you are managing a homogeneous ecosystem then it’s imperative that you use tools that natively support that system. Don’t try to use Linux tools to manage Windows and vice versa. If you do, you’re gonna have a bad time. If you are primarily deploying to Windows you should look at tools like Octopus Deploy or Build Master. If you’re managing a Linux ecosystem look into Chef, Puppet, or even Docker.
If you’re managing a mixed ecosystem where one OS was dominant, you should still use tools designed to support the dominant system. It may be worth the effort to see if your existing tools can also manage the subdominant system. In our case it’s not worth the effort so we have instead moved toward an “appliance” model for our Linux servers. What this means is instead of managing a bunch of code to deploy RabbitMQ to Linux, we’re instead creating VM Images for the Rabbit installation which we can hydrate at will. We have far fewer resources who know how to administer Linux so this model works better for us.
5. DevOps tools are in their infancy
DevOps tools are optimized for the problems their creators were facing. There are many more problems in the DevOps space than any of the dominant tools are capable of managing on their own.
For example, Chef wants to deploy a machine. It’s not primarily concerned about applications. The Chef model is to declare the state of the machine and then let Chef decide how to bring the machine to that state. This approach optimizes for horizontally scaling hundreds or thousands of identical nodes with very few commands. Awesome!
In our organization we see the world in terms of Applications–not machines. Our whole way of thinking about deployment is different than the way Chef looks at it. This isn’t a deficiency in Chef or in the way we look at the world, but when we started using Chef we weren’t aware of how fundamental that difference in perspective would actually be.
Because Chef looks at the world in terms of nodes, it has no built-in (or even recommended) solution for artifact and version management. We had to build that. We had to build solutions for managing cookbook versions, publishing artifact and cookbook versions into targeted environments, and forwarding changes to production to antecedent environments.
If you’re using Octopus (we’re migrating from Chef to Octopus) and looking at the world in terms of applications, you will have problems when you need to spin up new environments and whole machines with many applications pre-installed. Either way, you will have to build other tools to glue the off-the-shelf tools together.
(Aside: Though I am not personally a fan of Chef, I have heard of people using Chef to deploy their infrastructure and using Octopus to deploy applications.)
6. DevOps “best practices” are in their infancy
Chef likes to advertise “use Chef however you want! We’re flexible!” Great…. except Chef is complicated and I would like some guidance on how to use it! This isn’t so much a problem with Chef though–DevOps in general is a very young field so we don’t have the wealth of shared experience from which to draw generalized lessons. To the extent that there is guidance it’s basically cribbed from Software Engineering best practices and doesn’t always apply well.
Here are some of mine:
- Have a canary environment that rebuilds all machines and redeploys all software on a regularly scheduled basis. Use this environment to detect problems in your deployment tool chain early.
- Every developer should have an individual environment of their own to test deployments.
- Every team should have at least one environment for testing and/or UAT.
- Avoid “Standard Failures.” These are errors that occur often and either do not have a known solution or have a manual workaround. Identify the root cause of errors and address them. Incorporate manual workaround solutions into your automated solutions.
- Where possible, embed some sort of “health check” into your applications that you can invoke to have the application check it own configuration.
- Identify rollback strategies for your applications.
7. Developers will have to learn infrastructure
If you come from a development background you will have to learn about security, networking, hardware, virtual hardware, etc. This is the domain you are working in now. I’m still at the beginning of this process myself but I’m starting to see the size of how much I still have to learn. For example, if you’re deploying to the cloud you’ll have to learn the inner workings of your chosen cloud infrastructure.
8. Ops will have to learn development patterns and practices.
If you come from an Ops background you will have to learn Software Engineering patterns and practices. You are graduating from someone who writes the occasional script to someone who manages code. Writing some code that only has to be run once is easy. Writing code that has to work again and again and again as well as tolerate change is much, much harder. As the number of people, environments, and machines grow software engineering skills will become more and more important.
9. Don’t automate a bad process.
Consider this: Chef doesn’t provide a built-in way to define which artifacts should be deployed to which environments. To that end we built an “application versions” cookbook which contains a list of all applications, their version, and their artifact location. In order to start work a team must:
- Take a branch of the application versions cookbook.
- Edit the versions/artifact information.
- Upload the cookbook to Chef
- commit and push the changes back to github
- clone the chef-repo
- edit the affected environment to use the new version of the application versions cookbook.
- commit/push chef-repo
- upload the edited environment to Chef.
Does that sound like a good idea to you? It doesn’t to me–but it’s necessary if you’re going to use a Chef Cookbook as a source for environment application versions. Before you go and wrap some automation around this to make it “easier,” let’s challenge the basic assumption: should we maybe just store application versions by environment elsewhere? A json file on a network share would be easier than this.
When you automate a process (even to make it “easier”) you’re pouring a certain amount of lime over it. Be careful.
10. “Infrastructure as Code!” is not always a good idea.
Code != Artifacts != Configuration. The daily work of DevOps breaks down into basically three disciplines: Code, Configuration, and Artifact management. A change to one of these should not necessitate a change to the other. That means that Code, Configuration, and Artifacts should not live together in github.
Use a Package Manager for your artifacts. If you don’t know where to look check out Artifactory. It’s a versatile artifact repository that supports many different kinds of package managers. It’s API even understands version numbers and will let you identify and retrieve the “latest” version of your artifact. Let your CI server publish artifacts to your package manager and make it the canonical source for artifact retrieval.
Configuration should not be managed like code. Configuration data is any data required by applications to run. Examples are things like dns addresses, email addresses for notifications, database connection strings, api endpoints, etc.. Configuration data is just data about environments. Unlike code it does not need to be branched. It should be stored in some central repository and accessed directly by the deployment code.
The code that you use to execute your deployments is most emphatically and in every possible way code. This means it should be tested, stored in source control, subject to your company’s chosen branching strategies, built by a CI server, etc..
The “infrastructure as code” idea is a really great idea, but it applies only to the procedure of deploying hardware and software. It does not fit well with the metadata that describes which hardware and software should be deployed. Don’t use “infrastructure as code” as an execute to push square pegs into round holes.
As part of our Octopus Deploy migration effort we are writing a powershell module that we use to automatically bootstrap the Tentacle installation into Octopus. This involves maintaining metadata about machines and environments outside of Octopus. The reason we need this capability is to adhere to the “cattle vs. pets” approach to hardware. We want to be able to destroy and recreate our machines at will and have them show up again in Octopus ready to receive deployments.
Our initial implementation cycled through one machine at a time, installing Tentacle, registering it with Octopus (with the same security certificate so that Octopus recognizes it as the same machine), then moving on to the next machine. This is fine for small environments with few machines, but not awesome for larger environments with many machines. If it takes 2m to install Tentacle and I have 30 machines, I’m waiting an hour to be able to use the environment. With this problem in mind I decided to figure out how we could parallelize the boostrapping of machines in our Powershell module.
Start-Job
Start-Job is one of a family of Powershell functions created to support asynchrony. Other related functions are Get-Job
, Wait-Job
, Receive-Job
, and Remove-Job
. In it’s most basic form, Start-Job
accepts a script block as a parameter and executes it on a background thread.
# executes "dir" on a background thread.
$job = Start-Job -ScriptBlock { dir }
The job object returned by Start-Job
gives you useful information such as the job id, name, and current state. You can run Get-Job
to get a list of running jobs, Wait-Job
to wait on one or more jobs to complete, Receive-Job
to get the output of each job, and Remove-Job
to delist jobs in the current Powershell session.
Complexity
If that’s all there was to it, I wouldn’t be writing this blog post. I’d just tweet the link to the Start-Jobs
msdn page and call it done. My scenario is that I need to bootstrap machines using code defined in my Powershell module, but run those commands in a background process. I also need to collate and log the output of those processes as well as report on the succes/failure of each job.
When you call Start-Job
in Powershell it creates a new session in which currently loaded modules are not automatically loaded. If you have your powershell module in the $PsModulePath
you’re probably okay. However, there is a difference between the version of the module I’m currently working on and testing vs. the one I have on my machine for normal use.
Start-Job
has an additional parameter for a script block used to initialize the new Powershell session prior to executing your background process. The difficulty is that while you can pass arguments to the background process script block, you cannot pass arguments to the initialization script. Here’s how you make it all work.
Setup Code
<br /># Store the working module path in an environment variable so that the new powershell session can locate the correct version of the module.
# The environment variable will not persist beyond the current powershell session so we don't have to worry about poluting our machine state.
$env:OctobootModulePath = (get-module Octoboot).Path
$init = {
# When initializing the new session, use the -Force parameter in case a different version of the module is already loaded by a profile.
import-module $env:OctobootModulePath -Force
}
# create a parameterized script block
$scriptBlock = {
Param(
$computerName,
$environment,
$roles,
$userName,
$password,
$apiKey,
)
Install-Tentacle -computer $computerName `
-environment $environment `
-roles $roles `
-userName $userName -password $password `
-apiKey $apiKey
}
# I like to use an -Async switch on the controlling function. Debugging issues is easier in a synchronous context than in an async context. Making the async functionality optional is a win.
if ($async) {
$job = Start-Job `
-ScriptBlock $scriptBlock `
-InitializationScript $init `
-Name "Install Tentacle on $($computerName)" `
-ArgumentList @(
$computerName, $environment, $roles
$userName, $password,
$apiKey) -Debug:$debug
} else {
Install-Tentacle -computer $computerName `
-environment $environment `
-roles $roles `
-userName $userName -password $password `
-apiKey $apiKey
}
The above code is in a loop in the controlling powershell function. After I’ve kicked off all of the jobs I’m going to execute, I just need to wait on them to finish and collect their results.
Finalization Code
<br />if ($async) {
$jobs = get-job
$jobs | Wait-Job | Receive-Job
$jobs | foreach {
$job = $_
write-host "$($job.Id) - $($job.Name) - $($job.State)"
}
$jobs | remove-job
}
Since each individual job is now running in parallel, bootstrapping large environments doesn’t take much longer than bootstrapping smaller ones. The end result is that hour is now reduced to a few minutes.
DevOps is a relatively new space in the software engineering world. There are a smattering of tools to aid in the automation of application deployments, but precious little guidance with respect to patterns and practices for using the new tools. As a guy who loves leaning on principles this lack of attention to best practices leaves me feeling a bit uncomfortable. Since I’m leading a migration to Octopus Deploy, I thought I would share some of the decisions we’ve made.
This series of posts is an attempt to start a conversation about best practices. I want to be clear: We have not been applying these ideas long enough to know what all of the ramifications are. Your mileage may vary.
Posts in this series
1. Environments
2. Roles
3. Variables & Variable Sets
Variables & Variable Sets
Octopus Deploy allows you to modify your application’s configuration through the use of variables. You can define variables at the project level, or share variable values between projects through variable sets. If you have relatively little sharing of variables between projects you will likely prefer to create variables at the project level. My team manages over 50 different applications. Many of them are web services designed to support SOA. The net impact is that we have a lot of shared variables and for this reason we define variables exclusively through variable sets. This saves us time hunting for where a given variable is defined.
We use 2 kinds of variable sets
1. Global
2. Role based
Global variable sets define values that might be required across the company irrespective of any particular application, or that are more easily managed together. For example, we wish to capture metadata about environments. Octopus itself does not have a facility for tagging environments with arbitrary metadata. To satisfy this goal we created a variable set called “environment” in which we create variables to indicate values such as “owner” and “abbr”. We use these values to compose the values of other variables such as dns addresses or email addresses.
We also have some environments for which we do not create dns addresses for the sites. In these environments we need to install web applications with alternate ports. We keep a variable sets to define the port number we use for web applications in these environments since they must be unique across the web server.
The number of global variable sets should be as small as possible.
Role based variable sets are variables defined for the specific roles they target. If we have a role called heroes-iis
we will also have a variable set called heroes-iis
. Since we create roles on a per-deployed-application basis, this helps us keep roles, projects, and variable sets linked. If heroes-iis
as web service end points, this variable set may be included in some other project that depends on those end points.
Naming Conventions
It is important to have naming conventions for your variables. I highly recommend prefixing all variables in a variable set with the name of the variable set to avoid potential naming collisions. For example, If I have a variable set called heroes-iis
it will have variables with names like:
- heroes-iis.application-pool.name
- heroes-iis.application-pool.password
Define a Standard Structure for Similar Variable Sets
Once you get the rhythm of installing applications with Octopus, you will discover that similar kinds of applications have similar variable definition needs. You can save yourself a lot of time and Chrome tabs by establishing a variable set template that you use when creating a variable set for each kind of application you deploy. Here is our variable set template for web applications being deployed to iis:
Variable Set Name | Segment | Field | Variable Name | Notes |
---|---|---|---|---|
name-iis | application-pool | name | name-iis.application-pool.name | The name of the application pool |
username | name-iis.application-pool.username | the username the application pool runs under | ||
password | name-iis.application-pool.password | the password the application pool runs under | ||
host | name.host | This corresponds to the site name as registered in IIS. It does not include the protocol (http://, https://). It should be blank if the site is being deployed into an environment without a dns entry. | ||
site-name | name.site-name | This will be just the name of the web application in environments that do not have a dns entry. If the environment has a dns entry, it should resolve the host property. | ||
site-root | name.site-root | This is the url root for the site. It should include the protocol (http://, https://) as well as the port, and any additional routing information. | ||
endpoints | endpoint-name | name.endpoints.endpoint-name | A web service may expose one or more endpoints. These should have unique names. Their values should be defined with reference to the host and port variables. | connection-strings | cs-name | name.connection-strings.cs-name | The name of the connection string in the config file. |
Scope
Octopus Deploy allows you to scope variables by environment, role, or channel (as of 3.3). The scoping rules are as follows:
(environment1 OR environment2 OR ...) AND (role1 OR role2 OR ...) AND (channel1 OR channel2 OR ...)
I recommend that you scope variable values as broadly as possible. Use composed variable values where you can to minimize the number of variable values you have to maintain. For example:
<br />heroes-iis.connection-strings.heroes-db => "Server=#{environment.sql-server.url}; Database=#{heroes-db.database-name};"
heroes-db.database-name => #{HEROES_}#{environment.name}
environment.sql-server.url => http://sql-server.#{environment.name}.com
By using a composed variable value I don’t need to scope the connection string variable itself. Instead, I can confine scoping to environment.name
and satisfy the resolution of all of the descendant variables. This minimizes the number of variables I have to actively maintain as new environments are created.
DevOps is a relatively new space in the software engineering world. There are a smattering of tools to aid in the automation of application deployments, but precious little guidance with respect to patterns and practices for using the new tools. As a guy who loves leaning on principles this lack of attention to best practices leaves me feeling a bit uncomfortable. Since I’m leading a migration to Octopus Deploy, I thought I would share some of the decisions we’ve made.
This series of posts is an attempt to start a conversation about best practices. I want to be clear: We have not been applying these ideas long enough to know what all of the ramifications are. Your mileage may vary.
Posts in this series
1. Environments
2. Roles
3. Variables & Variable Sets
Roles
When you add machines into Octopus, you must specify environments and roles for that machine. For our purposes, environments were pretty easy to define. Roles however took some work. Here are the kinds of roles we defined.
Operating Systems
Example: windows, linux
This is pretty easy. We started with Linux
and Windows
for this type of role. I can see a day when we may need to additionally specify ubuntu-14
or 2k8-r2
. In the meantime, YAGNI.
Environment Types
Example: dev, uat, integration, staging, prod, support
Our environment naming convention for developer environments is dev-{first initial}{last name}
. For uat environments it’s uat-{team}
. There is only one of each integration, staging, production, and support environments. There are certain variables that are defined consistently across all dev
environments but may differ in uat
environments. For this reason we are applying the environment type as a role across all machines in the relevant environments.
Commands
Example: hero-db.migrator
This is a standalone role. There will only be one machine in each environment that will have this role. It’s purpose is to execute commands on some resource in the enviornment that should not be run multiple times or concurrently. A good example of this is an Entity Framework database migration. We choose one machine in an environment that database migrations can be run from.
Applications
Example: webapp-iis, topshelf-service
Each deployable application has its own role. Not every application gets installed on every machine in an environment. We use the -iis
affix for applications installed into IIS regardless of whether they’re sites or web services. We use the -service
affix for Windows Services. We do this because we sometimes have a family of applications that have the same name but target a different kind of application.
DevOps is a relatively new space in the software engineering world. There are a smattering of tools to aid in the automation of application deployments, but precious little guidance with respect to patterns and practices for using the new tools. As a guy who loves leaning on principles this lack of attention to best practices leaves me feeling a bit uncomfortable. Since I’m leading a migration to Octopus Deploy, I thought I would share some of the decisions we’ve made.
This series of posts is an attempt to start a conversation about best practices. I want to be clear: We have not been applying these ideas long enough to know what all of the ramifications are. Your mileage may vary.
Posts in this series
1. Environments
2. Roles
3. Variables & Variable Sets
Our Default Lifecycle
Before I begin, I should give you some background on our development ecosystem. Our Octopus Lifecycle looks like this:
dev => uat => integration => staging => prod => support
The Environments
Name | Convention | Purpose | Notes |
---|---|---|---|
dev | dev-{first initial}{last name} | The primary purpose of these environments is to test the deployment tooling itself. | We have 15 or so individual developer environments. Each developer gets their own environment with 2 servers (1 Linux, 1 Windows) and all of our 60 or so proprietary applications installed to it. |
uat | uat-{team} | These environments are used by teams to test their work. | We have 10 or so User Acceptance Testing environments. These are a little bit more fleshed out in terms of hardware. There are multiple web servers behind load balancers. The machines are beefier. These enviornments are usually owned by a single team, though they may sometimes be shared. |
integration | N/A | Dress rehearsal by Development for releases | The integration environment is much closer to production. When multiple teams are releasing their software during the same release window, integration gives us a rehearsal environment to make sure all of the work done by the various teams will work well together. |
staging | N/A | Dress rehearsal by Support for releases | Staging is exactly like integration except that it is not owned by the Development department. We have a team of people who are responsible for executing releases. This is their environment to verify that the steps development gave them will work. |
prod | N/A | Business Use | Prod is not managed by the deployment engineering team. We build the button that pushes to prod, but we do not push it. |
support | N/A | Rehearsal environment for support solutions | Support is a post-production environment that mirrors production. It allows support personnel to test and verify support tasks in a non-prod environment prior to running them in production. |