Related
At the end of a TeamCity build, it sends the artifacts back to the TC server. I don't need this to happen because my build script is pushing them to S3 where our web servers pick the payload up from. Is there a way I can disable them from being sent back, it's taking 1 min 30 seconds just for that step because the files are so large.
This might be most easily accomplished by clearing out the Artifact paths field in the General Settings of you build configuration. TeamCity should only be saving the things that are included there (in addition to a few internal reporting items.)
At work we currently use the following deployment strategy:
Run a batch script to clear out all Temporary ASP.NET files
Run a batch script that compiles every ASPX file into its own DLL (ASP.NET Web Site, not Web Application)
Copy each individually changed file (ASPX and DLL) to the appropriate folder on the live server.
Open up the Deployment Scripts folder, run each SQL script (table modifications, stored procs, etc.) manually on the production database.
Say a prayer before going to sleep (joking on this one, maybe)
Test first thing the next morning and hope for the best - fix bugs as they come up.
We have been bitten a few times in the past because someone will forget to run a script, or think they ran something but didn't, or overwrote a sproc related to some module because there are two files (one in a Sprocs folder and one in a [ModuleName]Related folder) or copied the wrong DLL (since they can have the same names with like a random alphanumeric number generated by .NET).
This seems highly inefficient to me - a lot of manual things and very error prone. It can sometimes take 2-3 or more hours for a developer to perform a deployment (we do them late at night, like around midnight) due to all the manual steps and remembering what files need to be copied, where they need to be copied, what scripts need to be run, making sure scripts are run in the right order, etc.
There has got to be an easier way than taking two hours to copy and paste individual ASPX pages, DLLs, images, stylesheets and the like and run some 30+ SQL scripts manually. We use SVN as our source control system (mainly just for update/commit though, we don't do branching) but have no unit tests or testing strategy. Is there some kind of tool that I can look into to help us make our deployments smoother?
I did not go through all of it, but the You're deploying it wrong series from Troy Hunt might be a good place to look at.
Points discussed in the series :
Config transforms
Build Automation
Continuous Integration
We have four stages before it can be deployed.
Development
QA
UAT
Production
We have build scripts (inside bamboo build server) running against QA and against UAT. Our DBA is the only person who can run create scripts against QA, UAT, and PROD. Anything that goes from QA -> UAT is like a test run deployment. UAT gets reverted by copying the production systems down again.
When we release into Production we just create a whole new site and point it at the UAT database and test that environmentally it is working fine. Then when this is working good we flick the 'switch' and point the production IIS record at the next site, and change the DB connection to point at Prod DB.
Because we are using a completely diff folder structure all of our files get copied up so there is no chance of missing one. Because we have had test runs of deployment into UAT we know we haven't missed a DB script (DB Scripts are combined into one generally). Because we have tested a shadow copy of the IIS website we know that environmentally it should work. We can then do all this set up during the day - and then do the final switch flicking at midnight or whenever - reducing the impact on devs.
tl;dr; Automated build and deploy; UAT system for test running deployment; Deployment during work hours; Flick switch/run DB update at midnight.
I am a developer for BuildMaster, a tool which can very easily automate the steps you have outlined above, and we have a limited version free for a team of 5 developers.
Most of your pain points will disappear the moment you set up the deployment automation - mainly the batch script execution and the file-by-file copying. Once you're fully automated, you can even schedule the deployment for night time and only have to worry about it if there's an error in the process (you can set up a notifier for a failed build).
On the database side, you can integrate your database with BuildMaster as well and if you upload the scripts into the tool it will keep track of which ones were run against which database.
To see how to set up a simple web application deployment plan, you can run one of the example applications included. You can also check out: http://inedo.com/support/tutorials/lunchmaster/part-1 to see how to create one yourself - it's slightly outdated since we've made it even easier to get started out-of-the-box but the main concepts are the same.
Please see this blog post and associated talk by Scott Hanselman titled "Web Deployment Made Awesome"
Blog
Video
As for SQL Deployment, you might want to consider one of the following:
RikMigrations
Migrator.NET
FluentMigrator
Mantee Introduction & Source
Have a User Acceptance Testing (UAT) environment which is completely isolated from your development environment, and only accessible to the UAT manager.
Setup a UAT build which you can manually trigger upon each release, when triggered this should send all your deployment files as well as a deployment checklist to the UAT manager, who will redeploy all files to the UAT environment, and run any database upgrade scripts.
Once the applications users and testers have signed off the UAT release, the UAT manager can be authorised to deploy to the PRODUCTION environment using the exact same procedure and checklists as the UAT release. This will guarantee that you never miss any deployment steps, and test the deployment process prior to moving it into production.
Caveats- I'm in an environment where we can't use MSI, batch, etc for the final deployment
Things that helped:
A build server that does the full compilation on a build server and runs all unit tests and integration tests. Why find out you have something in an aspx page that doesn't compile on deployment night? (I admit your Q doesn't make it clear if compilation is happening on deployment night)
I have a page that administrators can reach that exercises environment and deployment failure points, e.g. connect to db, connect to reporting services, send an email, read and write to the temp folder.
Also, put all the things that the administrator needs to change into a file external from web.config. The connection string and app settings sections natively support a way to do this (i.e. don't reinvent the web.config system just to create a separate file)
Here is an article on how to do better integration tests: http://suburbandestiny.com/Tech/?p=601 There is a ton of good literature how to do unit tests, but often if you app already exists, you will have to refactor until unit testing becomes possible. If that isn't an option, then don't be a purist and put together some integration tests that are fast and repeatable as possible.
Keep your dependencies in bin instead of GAC, since it's easier to tell an administrator to copy files than it is to teach them to administer the GAC.
The process of creating a new build and releasing it to production is a critical step in the SDLC but it is often left as an afterthought and varies greatly from one company to the next.
I'm hoping people will share improvements they have made to this process in their organisation so we can all takes steps to 'reduce the pain'.
So the question is, specify one painful/time consuming part of your release process and what did you do to improve it?
My example: at a previous employer all developers made database changes on one common development database. Then when it came to release time, we used Redgate's SQL Compare to generate a huge script from the differences between the Dev and QA databases.
This works reasonably well but the problems with this approach are:-
ALL changes in the Dev database are included, some of which may still be 'works in progress'.
Sometimes developers made conflicting changes (that were not noticed until the release was in production)
It was a time consuming and manual process to create and validate the script (by validate I mean, try to weed out issues like problem 1 and 2).
When there were problems with the script (eg the order in which things were run such as creating a record which relies on a foreign key record which is in the script but not yet run) it took time to 'tweak' it so it ran smoothly.
It's not an ideal scenario for Continuous Integration.
So the solution was:-
Enforce a policy of all changes to the database must be scripted.
A naming convention was important for ensuring the correct running order of the scripts.
Create/Use a tool to run the scripts at release time.
Developers had their own copy of the database do develop against (so there was no more 'stepping on each others toes')
The next release after we started this process was much faster with fewer problems, indeed the only problems found were due to people 'breaking the rules', eg not creating a script.
Once the issues with releasing to QA were fixed, when it came time to release to production it was very smooth.
We applied a few other changes (like introducing CI) but this was the most significant, overall we reduced release time from around 3 hours down to a max of 10-15 minutes.
We've done a few things over the past year or so to improve our build process.
Fully automated and complete build. We've always had a nightly "build" but we found that there are different definitions for what constitutes a build. Some would consider it compiling, usually people include unit tests, and sometimes other things. We clarified internally that our automated build literally does everything required to go from source control to what we deliver to the customer. The more we automated various parts, the better the process is and less we have to do manually when it's time to release (and less worries about forgetting something). For example, our build version stamps everything with svn revision number, compiles the various application parts done in a few different languages, runs unit tests, copies the compile outputs to appropriate directories for creating our installer, creates the actual installer, copies the installer to our test network, runs the installer on the test machines, and verifies the new version was properly installed.
Delay between code complete and release. Over time we've gradually increased the amount of delay between when we finish coding for a particular release and when that release gets to customers. This provides more dedicated time for testers to test a product that isn't changing much and produces more stable production releases. Source control branch/merge is very important here so the dev team can work on the next version while testers are still working on the last release.
Branch owner. Once we've branched our code to create a release branch and then continued working on trunk for the following release, we assign a single rotating release branch owner that is responsible for verifying all fixes applied to the branch. Every single check-in, regardless of size, must be reviewed by two devs.
We were already using TeamCity (an excellent continuous integration tool) to do our builds, which included unit tests. There were three big improvements were mentioning:
1) Install kit and one-click UAT deployments
We packaged our app as an install kit using NSIS (not an MSI, which was so much more complicated and unnecessary for our needs). This install kit did everything necessary, like stop IIS, copy the files, put configuration files in the right places, restart IIS, etc. We then created a TeamCity build configuration which ran that install kit remotely on the test server using psexec.
This allowed our testers to do UAT deployments themselves, as long as they didn't contain database changes - but those were much rarer than code changes.
Production deployments were, of course, more involved and we couldn't automate them this much, but we still used the same install kit, which helped to ensure consistency between UAT and production. If anything was missing or not copied to the right place it was usually picked up in UAT.
2) Automating database deployments
Deploying database changes was a big problem as well. We were already scripting all DB changes, but there were still problems in knowing which scripts were already run and which still needed to be run and in what order. We looked at several tools for this, but ended up rolling our own.
DB scripts were organised in a directory structure by the release number. In addition to the scripts developers were required to add the filename of a script to a text file, one filename per line, which specified the correct order. We wrote a command-line tool which processed this file and executed the scripts against a given DB. It also recorded which scripts it had run (and when) in a special table in the DB and next time it did not run those again. This means that a developer could simply add a DB script, add its name to the text file and run the tool against the UAT DB without running around asking others what scripts they last ran. We used the same tool in production, but of course it was only run once per release.
The extra step that really made this work well is running the DB deployment as part of the build. Our unit tests ran against a real DB (a very small one, with minimal data). The build script would restore a backup of the DB from the previous release and then run all the scripts for the current release and take a new backup. (In practice it was a little more complicated, because we also had patch releases and the backup was only done for full releases, but the tool was smart enough to handle that.) This ensured that the DB scripts were tested together at every build and if developers made conflicting schema changes it would be picked up quickly.
The only manual steps were at release time: we incremented the release number on the build server and copied the "current DB" backup to make it the "last release" backup. Apart from that we no longer had to worry about the DB used by the build. The UAT database still occasionally had to be restored from backup (eg. since the system couldn't undo the changes for a deleted DB script), but that was fairly rare.
3) Branching for a release
It sounds basic and almost not worth mentioning, yet we weren't doing this to begin with. Merging back changes can certainly be a pain, but not as much of a pain as having a single codebase for today's release and next month's! We also got the person who made the most changes on the release branches to do the merge, which served to remind everyone to keep their release branch commits to an absolute minimum.
Automate your release process whereever possible.
As others have hinted, use different levels of build "depth". For instance a developer build could make all binaries for runnning your product on the dev machine, directly from the repository while an installer build could assemble everything for installation on a new machine.
This could include
binaries,
JAR/WAR archives,
default configuration files,
database scheme installation scripts,
database migration scripts,
OS configuration scripts,
man/hlp pages,
HTML documentation,
PDF documentation
and so on. The installer build can stuff all this into an installable package (InstallShield, ZIP, RPM or whatever) and even build the CD ISOs for physical distribution.
The output of the installer build is what is typically handed over to the test department. Whatever is not included in the installation package (patch on top of the installation...) is a bug. Challenge your devs to deliver a fault free installation procedure.
Automated single step build. The ant build script edits all the installer configuration files, program files that need changed ( versioning) and then builds. No intervention required.
There is still a script run to generate the installers when it's done, but we will eliminate that.
The CD artwork is versioned manually; that needs fixed too.
Agree with previous comments.
Here is what has evolved where I work. This current process has eliminated the 'gotchas' that you've described in your question.
We use ant to pull code from svn (by tag version) and pull in dependencies and build the project (and at times, also to deploy).
Same ant script (passing params) is used for each env (dev, integration, test, prod).
Project process
Capturing requirements as user 'stories' (helps avoid quibbling over an interpretation of a requirement, when phrased as a meaningful user interaction with the product)
following an Agile principles so that each iteration of the project (2 wks) results in demo of current functionality and a releasable, if limited, product
manage release stories throughout the project to understand what is in and out of scope (and prevent confusion abut last minute fixes)
(repeat of previous response) Code freeze, then only test (no added features)
Dev process
unit tests
code checkins
scheduled automated builds (cruise control, for example)
complete a build/deploy to an integration environment, and runs smoke test
tag the code and communicate to team (for testing and release planning)
Test process
functional testing (selenium, for example)
executing test plans and functional scenarios
One person manages the release process, and ensures everyone complies. Additionally all releases are reviewed a week before launch. Releases are only approved if there are:
Release Process
Approve release for a specific date/time
Review release/rollback plan
run ant with 'production deployment' parameter
execute DB tasks (if any) (also, these scripts can be version and tagged for production)
execute other system changes / configs
communicate changes
I don't know or practice SDLC, but for me, these tools have been indispensible in achieving smooth releases:
Maven for build, with Nexus local repository manager
Hudson for continuous integration, release builds, SCM tagging and build promotion
Sonar for quality metrics.
Tracking database changes to development db schema and managing updates to qa and release via DbMaintain and LiquiBase
On a project where I work we were using Doctrine's (PHP ORM) migrations to upgrade and downgrade the database. We had all manner of problems as the generated models no longer matched with the database schema causing the migrations to completely fail half way.
In the end we decided to write our own super basic version of the same thing - nothing fancy, just up's and down's that execute SQL. Anyway it worked out great (so far - touch wood). Although we were reinventing the wheel slightly by writing our own, the fact that the focus was on keeping it simple meant that we have far less problems. Now a release is a cinch.
I guess the moral of the story here is that it is sometimes OK to reinvent the wheel some times as long as you are doing so for a good reason.
We have multiple BizTalk 2006 application servers, and I find it almost impossible to keep the versions of our projects in sync on them. It's a tedious process of deploying the MSI packages, importing them, matching up files in the GAC, deploying some registry changes, and if one step is missed or somebody deployed an updated copy of a DLL directly to one server and not another, there's no easy way to tell.
How do others ensure that copies of software between the two servers are the same version?
Some Background:
Our environment has two (non-clustered) BizTalk front end servers and a separate database back-end. Until recently, though we had both front-ends configured, the host instances were stopped on the second server because of some troubleshooting. They've been disabled for a few months, and we're deployed some updated code in the meantime.
This morning, I did a folder diff on the GAC, as well as the folder that holds the local disk copy of the DLLs for our deployed project (C:\OurProject\ on both servers), and everything matched - same file sizes, same timestamps. However, once I turned on the second set of services, it became obvious that Server2 was using an old version of the project DLL - of the next three files processed, two had normal results and one was clearly out of date.
Please help me avoid an aneurysm.
One thing you may want to look into is the BizTalk Deployment Framework.
We are currently building up a new environment with BizTalk 2009 and I started out with a set of MSBuild scripts that handle exporting sources from SubVersion, building and deploying assemblies using BTSTask.
Of course BTSTask lacks a lot of functionality (start/stop applications) but at least for BizTalk 2006 there is BTSControl.
We use an automated build script whose ultimate end is an MSI with binding files for Dev/Stage/Prod. All released binding files are stored on a share and used to load the BizTalk server by hand. First the App is stopped, MSI's executed on both servers and then MSI imported. During the import, we specify the environment for the bindings and voila. We've had no issues with loss of synch.
So, I'd suggest taking all of your latest MSIs and re-execute them on the servers where you have differences. Otherwise, just try to put a process in place to create a repeatable load process by hand.
I am looking for different techniques/tools you use to deploy an ASP.NET web application project (NOT ASP.NET web site) to production?
I am particularly interested of the workflow happening between the time your Continuous Integration Build server drops the binaries at some location and the time the first user request hits these binaries.
Are you using some specific tools or just XCOPY? How is the application packaged (ZIP, MSI, ...)?
When an application is deployed for the first time how do you setup the App Pool and Virtual Directory (do you create them manually or with some tool)?
When a static resource changes (CSS, JS or image file) do you redeploy the whole application or only the modified resource? How about when an assembly/ASPX page changes?
Do you keep track of all deployed versions for a given application and in case something goes wrong do you have procedures of restoring the application to a previous known working state?
Feel free to complete the previous list.
And here's what we use to deploy our ASP.NET applications:
We add a Web Deployment Project to the solution and set it up to build the ASP.NET web application
We add a Setup Project (NOT Web Setup Project) to the solution and set it to take the output of the Web Deployment Project
We add a custom install action and in the OnInstall event we run a custom build .NET assembly that creates an App Pool and a Virtual Directory in IIS using System.DirectoryServices.DirectoryEntry (This task is performed only the first time an application is deployed). We support multiple Web Sites in IIS, Authentication for Virtual Directories and setting identities for App Pools.
We add a custom task in TFS to build the Setup Project (TFS does not support Setup Projects so we had to use devenv.exe to build the MSI)
The MSI is installed on the live server (if there's a previous version of the MSI it is first uninstalled)
We have all of our code deployed in MSIs using Setup Factory. If something has to change we redeploy the entire solution. This sounds like overkill for a css file, but it absolutely keeps all environments in sync, and we know exactly what is in production (we deploy to all test and uat environments the same way).
We do rolling deployment to the live servers, so we don't use installer projects; we have something more like CI:
"live" build-server builds from the approved source (not the "HEAD" of the repo)
(after it has taken a backup ;-p)
robocopy publishes to a staging server ("live", but not in the F5 cluster)
final validation done on the staging server, often with "hosts" hacks to emulate the entire thing as closely as possible
robocopy /L is used automatically to distribute a list of the changes in the next "push", to alert of any goofs
as part of a scheduled process, the cluster is cycled, deploying to the nodes in the cluster via robocopy (while they are out of the cluster)
robocopy automatically ensures that only changes are deployed.
Re the App Pool etc; I would love this to be automated (see this question), but at the moment it is manual. I really want to change that, though.
(it probably helps that we have our own data-centre and server-farm "on-site", so we don't have to cross many hurdles)
Website
Deployer:
http://www.codeproject.com/KB/install/deployer.aspx
I publish website to a local folder, zip it, then upload it over FTP. Deployer on server then extracts zip, replaces config values (in Web.Config and other files), and that's it.
Of course for first run you need to connect to the server and setup IIS WebSite, database, but after that publishing updates is piece of cake.
Database
For keeping databases in sync I use http://www.red-gate.com/products/sql-development/sql-compare/
If server is behind bunch of routers and you can't directly connect (which is requirement of SQL Compare), use https://secure.logmein.com/products/hamachi2/ to create VPN.
I deploy mostly ASP.NET apps to Linux servers and redeploy everything for even the smallest change. Here is my standard workflow:
I use a source code repository (like Subversion)
On the server, I have a bash script that does the following:
Checks out the latest code
Does a build (creates the DLLs)
Filters the files down to the essentials (removes code files for example)
Backs up the database
Deploys the files to the web server in a directory named with the current date
Updates the database if a new schema is included in the deployment
Makes the new installation the default one so it will be served with the next hit
Checkout is done with the command-line version of Subversion and building is done with xbuild (msbuild work-alike from the Mono project). Most of the magic is done in ReleaseIt.
On my dev server I essentially have continuous integration but on the production side I actually SSH into the server and initiate the deployment manually by running the script. My script is cleverly called 'deploy' so that is what I type at the bash prompt. I am very creative. Not.
In production, I have to type 'deploy' twice: once to check-out, build, and deploy to a dated directory and once to make that directory the default instance. Since the directories are dated, I can revert to any previous deployment simply by typing 'deploy' from within the relevant directory.
Initial deployment takes a couple of minutes and reversion to a prior version takes a few seconds.
It has been a nice solution for me and relies only on the three command-line utilities (svn, xbuild, and releaseit), the DB client, SSH, and Bash.
I really need to update the copy of ReleaseIt on CodePlex sometime:
http://releaseit.codeplex.com/
Simple XCopy for ASP.NET. Zip it up, sftp to the server, extract into the right location. For the first deployment, manual set up of IIS
Answering your questions:
XCopy
Manually
For static resources, we only deploy the changed resource.
For DLL's we deploy the changed DLL and ASPX pages.
Yes, and yes.
Keeping it nice and simple has saved us alot of headaches so far.
Are you using some specific tools or just XCOPY? How is the application packaged (ZIP, MSI, ...)?
As a developer for BuildMaster, this is naturally what I use. All applications are built and packaged within the tool as artifacts, which are stored internally as ZIP files.
When an application is deployed for the first time how do you setup the App Pool and Virtual Directory (do you create them manually or with some tool)?
Manually - we create a change control within the tool that reminds us the exact steps to perform in future environments as the application moves through its testing environments. This could also be automated with a simple PowerShell script, but we do not add new applications very often so it's just as easy to spend the 1 minute it takes to create the site manually.
When a static resource changes (CSS, JS or image file) do you redeploy the whole application or only the modified resource? How about when an assembly/ASPX page changes?
By default, the process of deploying artifacts is set-up such that only files that are modified are transferred to the target server - this includes everything from CSS files, JavaScript files, ASPX pages, and linked assemblies.
Do you keep track of all deployed versions for a given application and in case something goes wrong do you have procedures of restoring the application to a previous known working state?
Yes, BuildMaster handles all of this for us. Restoring is mostly as simple as re-executing an old build promotion, but sometimes database changes need to be manually restored, and data loss can occur. The basic rollback process is detailed here: http://inedo.com/support/tutorials/performing-a-deployment-rollback-with-buildmaster
web setup/install projects - so you can easily uninstall it if something goes wrong
Unfold is a capistrano-like deployment solution I wrote for .net applications. It is what we use on all of our projects and it's a very flexible solution. It solves most of the typical problems for .net applications as explained in this blog post by Rob Conery.
it comes with a good "default" behavior, in the sense that it does a lot of standard stuff for you: getting the code from source control, building, creating the application pool, setting up IIS, etc
releases based on what's in source control
it has task hooks, so the default behaviour can be easily extended or altered
it has rollback
it's all powershell, so there aren't any external dependencies
it uses powershell remoting to access remote machines
Here's an introduction and some other blog posts.
So to answer the questions above:
How is the application packaged (ZIP, MSI, ...)?
Git (or another scm) is the default way to get the application on the target machine. Alternatively you can perform a local build and copy the result over the Powereshell remoting connection
When an application is deployed for the first time how do you setup the App Pool and Virtual Directory (do you create them manually or with some tool)?
Unfold configures the application pool and website application using Powershell's WebAdministration Module. It allows us (and you) to modify any aspect of the application pool or website
When a static resource changes (CSS, JS or image file) do you redeploy the whole application or only the modified resource? How about when an assembly/ASPX page changes?
Yes unfold does this, any deploy is installed next to the others. That way we can easily rollback
when somehting goes wrong. It also allows us to easily trace back a deployed version to
a source control revision.
Do you keep track of all deployed versions for a given application?
Yes, unfold keeps old versions around. Not all versions, but a number of versions. It makes rolling back almost trivial.
We've been improving our release process for the past year and now we've got it down pat. I'm using Jenkins to manage all of our automated builds and releases, but I'm sure you could use TeamCity or CruiseControl.
So upon checkin, our "normal" build does the following:
Jenkins does a SVN update to fetch the latest version of the code
A NuGet package restore is done running against our own local NuGet repository
The application is compiled using MsBuild. Setting this up is an adventure, because you need to install the correct MsBuild and then the ASP.NET and MVC dll's on your build box. (As a side note, when I had <MvcBuildViews>true</MvcBuildViews> entered in my .csproj files to compile the views, msbuild was randomly crashing, so I had to disable it)
Once the code is compiled the unit tests are run (I'm using nunit for this, but you can use anything you want)
If all the unit tests pass, I stop the IIS app pool, deploy the app locally (just a few basic XCOPY commands to copy over the necessary files) and then restart IIS (I've had problems with IIS locking files, and this solved it)
I have separate web.config files for each environment; dev, uat, prod. (I tried using the web transformation stuff with little success). So the right web.config file is also copied across
I then use PhantomJS to execute a bunch of UI tests. It also takes a bunch of screenshots at different resolutions (mobile, desktop) and stamps each screenshot with some information (page title, resolution). Jenkins has great support for handling these screenshots and they are saved as part of the build
Once the integration UI tests pass the build is successful
If someone clicks "Deploy to UAT":
If the last build was successful, Jenkins does another SVN update
The application is compiled using a RELEASE configuration
A "www" directory is created and the application is copied into it
I then use winscp to synchronise the filesystem between the build box and UAT
I send a HTTP request to the UAT server and make sure I get back a 200
This revision is tagged in SVN as UAT-datetime
If we've got this far, build is successful!
When we click "Deploy to Prod":
The user selects a UAT Tag that was previously created
The tag is "switched" to
Code is compiled and synced with Prod server
Http request to Prod server
This revision is tagged in SVN as Prod-datetime
The release is zipped and stored
All up a full build to production takes about 30 secs which I'm very, very happy with.
Upsides to this solution:
It's fast
Unit tests should catch logic errors
When a UI bug gets into production, the screenshots will hopefully show what revision # caused the it
UAT and Prod are kept in sync
Jenkins shows you a great release history to UAT and Prod with all of the commit messages
UAT and Prod releases are all tagged automatically
You can see when releases happen and who did them
The main downsides to this solution are:
Whenever you do a release to Prod you need to do a release to UAT. This was a conscious decision we made because we wanted to always ensure that UAT is always up to date with Prod. Still, it's a pain.
There's quite a few configuration files floating around. I've attempted to have it all in Jenkins, but there's a few support batch files needed as part of the process. (These are also checked in).
DB upgrade and downgrade scripts are part of the app and run at app startup. It works (mostly), but it's a pain.
I'd love to hear any other possible improvements!
Back in 2009, where this answer hails from, we used CruiseControl.net for our Continuous Integration builds, which also outputted Release Media.
From there we used Smart Sync software to compare against a production server that was out of the load balanced pool, and moved the changes up.
Finally, after validating the release, we ran a DOS script that primarily used RoboCopy to sync the code over to the live servers, stopping/starting IIS as it went.
At the last company I worked for we used to deploy using an rSync batch file to upload only the changes since the last upload. The beauty of rSync is that you can add exclude lists to exclude specific files or filename patterns. So excluding all of our .cs files, solution and project files is really easy, for instance.
We were using TortoiseSVN for version control, and so it was nice to be able to write in several SVN commands to accomplish the following:
First off, check the user has the latest revision. If not, either prompt them to update or run the update right there and then.
Download a text file from the server called "synclog.txt" that details who the SVN user is, what revision number they are uploading and the date and time of the upload. Append a new line for the current upload and then send it back to the server along with the changed files. This makes it extremely easy to find out what version of the site to roll back to on the off chance that an upload causes problems.
In addition to this there is a second batch file that just checks for file differences on the live server. This can highlight the common problem where someone would upload but not commit their changes to SVN. Combined with the sync log mentioned above we could find out who the likely culprit was and ask them to commit their work.
And lastly, rSync allows you to take a backup of the files that were replaced during the upload. We had it move them into a backup folder So if you suddenly realised that some of the files should not have been overwritten, you can find the last backup up version of every file in that folder.
While the solution felt a little clunky at the time I have since come to appreciate it a whole lot more when working in environments where the upload method is a lot less elegant or easy (remote desktop, copy and paste the entire site, for instance).
I'd recommend NOT just overwriting existing application files but instead create a directory per version and repointing the IIS application to the new path.
This has several benefits:
Quick to revert if needed
No need to stop IIS or the app pool to avoid locking issues
No risk of old files causing problems
More or less zero downtime (usually just a pause at the new appdomain initialises)
The only issue we've had is resources being cached if you don't restart the app pool and rely on the automatic appdomain switch.