We are currently running multiple branches within TFS 2013:
Development
Release
Trunk
etc.
Each branch has an automated test build. The purpose of these builds is to publish the latest code (and database changes) for each branch to our test server for manual testing. We maintain a separate IIS application for each branch so that different versions can be tested simultaneously.
Each branch has unique .csproj, .xml profile and web.config settings which control which application and database each branch deploys to, which works great.
However when we merge from one branch to another, auto-merge replaces these branch specific settings causing us to deploy our test builds to the wrong URL and database.
I am aware that auto-merge can be completely disabled in Visual Studio settings but is there a way we can continue to use auto-merge except for specific files?
Any suggestions welcome
This is a well known problem with any code promotion model. I would suggest that you move instead to a binary promotion model.
http://nakedalm.com/building-release-pipeline-release-management-visual-studio-2013/
With a binary promotion model you build the bits once with replaceable parameters and create or use a tool that orchestrates the release and feeds it with the correct variables for each environment.
http://nakedalm.com/create-release-management-pipeline-professional-developers/
I recently built out an end to end scenario using VSO. You can see in the post the passing of the values for each environment.
The software industry is moving towards continuous delivery and continuous integration. Neither of which is possible in a code promotion model.
You could disable auto conflict resolving during merge, but in this case you have to resolve these conflict manually (ignore)
Options -> Source Control -> Visual Studio Team Foundation Server
Attempt to automatically resolve conflicts when they are generated
or you could use tf.exe for merge:
tf.exe merge /discard CS
use a changeset which will be always discard from merge.
Related
I have been working with TeamCity, Jenkins and Bamboo my last 8years. Latest 2years very involved in ContinuousIntegration factory setup and mainteinance on my team with very good results, giving me a lot of habits about how to deal with builds, artifacts and pipelines.
Now, i'm on a new company, new team, new CI, TFS2015, first time for me.
Just one month before I arrive to this new team they were on TFS2012, with XAMLs, so I took the migration to vNext builds.
At first look, I found on vNext builds the classic build definition, i mean adding steps as a single task to the build, instead of monolithic XAML file.
But with the time, i was trying to create more complex builds, like TeamCity build chains, but this is not possible, strike one...
Then, I was trying to deal with multiple branches, one continuous build for each branch (we are on TFSVC), create packages from each branch, and I found I was duplicating my builds just to change repository paths and a few details, so I moved builds to templates trying to reuse build definitions, introducing variables to generates paths(for repos and branches) and versions and expecting to change only in one place the build and having this changes reflected in all builds derived from templates... but that's was not the case... :
variables are not accepted everywhere, like in repo paths.
can't change templates after created them, just replace them, and builds created from templates are not affected after template is changed.
strike two ?.....
I'm wondering if maybe I'm doing things wrong with TFS, maybe this is a different system and I can't do things like in other CIs.
Any advice in how to approach TFS to have a good, dynamic and reusable set of builds ??
here isn’t the feature that the changes of a build definition or template affects other existing build definitions.
If the build steps of build definitions are the same for each branch, you just need a build definition and add filters for each branch (Triggers > Continuous integrations (CI), after that it will uses corresponding source to build for CI build, for example, develop branch changes > Trigger a build automatically with develop branch source.
On the other hand, you can change branch and source version when queue build manually or through REST API.
If the build steps are different for each branch, you need to modify the definitions independently for detail requirements.
Our TFS Source Control is setup like this:
TFS Project
Solutions
WebApp1Solution
WebApp2Solution
Libraries
SharedLibrary1
Web Applications
WebApp1
WebApp2
When I open solution WebApp1Solution, it loads my workspace with projects from Libraries and Web Applications folder.
I would like to suspend my current workspace environment for both web apps and all shared libraries and start working on a specific version of this code from an earlier label (our last deployment). We do not have any branching as of yet.
From what I've read, I think these are my options:
1. Shelve pending changes on the solution and get the specific label version of the solution. What would happen when I check any changes back in?
2. Create a branch in Source Control from the specific label version. Would I have to create the branch on all of the folders in the different locations? How would I switch back and forth between my current code and new branch code?
Any help or advice will be greatly appreciated! Thanks!
Just create a branch at your TFS Project folder into a development folder. You should actually be doing this normally for all of your development. Your mainline branch should never contain anything except for your production deployable codebase. You should always be able to get latest of your mainline and build and run as the production end product.
All developers/teams of developers should have their own TFS folder where they can branch from mainline, make changes, test, etc... then once it has passed all approvals, merge it to mainline and destroy your devline.
As for "switching" your branch is a separate folder from your mainline. You can open both at the same time in different instances of Visual Studio, just like opening two solutions at once. Branching creates a copy of the folder/solution/project structure from the point of your branch. Then when you merge it back it reports differences and attempts to merge. If there are conflicts you will be prompted to review and resolve them. Most of the time TFS does a good job at auto-resolving, but I always carefully review all conflicts before selecting a method of resolution.
I have the following scenario:
I have a Dev and Stage branch
Code is checked into Dev, and at some point those changesets from Dev are merged into Stage
No direct edits/check-ins happen to stage (beyond the merges from dev)
What I want to do is identify/verify that no new code has been introduced directly into the stage branch which did not originate from a changeset in Dev.
It does not appear possible to secure Stage (from a source control perspective) to only allow check-ins of merges from dev (since all operations appear to be merge,edit after I do the merge), so what I thought I could do is create a TFS Check in policy which attempts to compare the merged changes to their source.
I have created a class which derives from PolicyBase and have over-ridden the Evaluate method. In the Evaluate method, I spin through this.PendingCheckin.PendingChanges.CheckedPendingChanges and check the .IsMerge property and .ServerItem property to see if the item in question is a merge and is destined for my Stage branch.
When I spin through each change however, I don't see a way to track its lineage or access the parent from where it might have originated (ie: the Dev branch version #).
How do I only allow merges, but no direct edits on my Stage branch?
The only way to achieve this in TFVC is for you to take ownership of all merges. If you do the merge then you have the control. Many organizations with these sorts of draconian policies create an automated script that does the merge and rejects if there are any conflicts. That way the developer is forced to pull from Staging to Dev and resolve all conflicts before requesting the merge through a web portal. If the merge is successful it was conducted on a server by a single service account and no coder required permissions to Staging of any sort.
This is however dysfunctional. Either you don't trust your coders, or they are not competent enough to work in your code base.
Another method might be to move to Git in TFS for source control. In Git merges are non-editable and are performed as a single action rather than the merge/edit you get in TFVC.
TeamCity appears to store the definitions for builds, projects, templates etc as XML internally.
This is exposed in the "Administration > Audit" view where you can see diffs that people made to individual configurations, at URLs like http://teamcityserver/admin/settingsDiffView.html?id=project:project10&versionBefore=8&versionAfter=9&actionId=3151
I'd like to manage a TeamCity setup partially from outside the web interface - e.g. for example keep the build definitions in version control and perhaps programmatically generate them.
Is there any way I can directly upload definitions in this format (or any similar alternative)? I'm aware that there are various APIs and extension points to TeamCity but haven't managed to find any that gives direct access to anything like this.
I can live with the format changing with TeamCity versions if necessary - it would be a reasonable price to pay for the other benefits.
For TeamCity 9.x and newer
As reported by Ganesh in the comments to this answer, an option was added in 9.x that supports changes and versioning through Source Code Management (SCM) tools. Please see his answer for 9.x and beyond.
For TeamCity 8.x and older
It might not be the "approved" way, but you can edit the project files on disk, and those changes will appear in your build configs. I have successfully edited them outside of the Web UI after they were created.
So, you could probably open that folder up as a restricted network share or set up ssh.
You'll find it at $TeamCityData/config/projects/ and then they are stored in subfolders such as $projectName/buildTypes/$buildFile.xml
An example is:
E:\TeamCityData\config\projects\CSandbox\buildTypes\CSandbox_Project1TrunkBuildUnitTest.xml
TeamCity 9 adds a new "Versioned Settings" feature which keeps these XML files under version control and allows changes to be made via the VCS.
In TeamCity 9.0 this can be git or mercurial, and the upcoming TeamCity 9.1 will add support for Perforce and Subversion.
I've been using it with git for a few months and it works quite nicely in practice.
I sometimes have trouble persuading TeamCity to notice changes coming in from the VCS - particularly when deleting projects - but otherwise it's been really useful for standardising configuration and spinning up new job chains quickly.
Another slight annoyance is that you can't configure the location within the repository that the settings come from - it's always .teamcity in the root - so I've had to use multiple branches or repositories to manage multiple TeamCity servers.
I am setting up a CruiseControl.Net server. So far, it only builds a project (.Net website), and I kind-of know how to set up unit testing, code coverage, etc in the future.
What I will need to have soon is this:
The developers commit changes to SVN continually, thus CCNet builds often.
CCNet will publish the latest version to the development server, as soon as a commit is validated (with unit tests etc).
The project manager validates a specific version, in order to publish it to the pre-production server, and create a SVN tag from this revision.
The last point is where my problem lies: how exactly can I set up things so the project manager can, for instance, browse to the CCNet web dashboard, select a previous specific build, and says "this is the build I want to publish" ?
I believe that my thinking is flawed somewhere, but I can't put my finger on it. Maybe CCNet is not the right place to do these manipulations ?
In my mind, I can create a SVN tag using CCNet, and mostly work from the trunk, but maybe I can't ? Maybe it's the other way around, and I should add a CCNet project every time a tag is created under SVN ?
The final goal is that I want to automate the publication process: zip creation (for archiving), web.config modification (using Nant for instance), and website publication (using FTP).
In all these steps, I want to limit the manual intervention to the maximum. If I can avoid to add a new project to CCNet every time a tag or branch is created in SVN, that would be awesome.
Thanks for your help, and sorry if it's not very easy to read, but it's not very clear in my head either...
Since you can create any task, you should be able to achieve the goal, though unfortunately not out-of-the-box.
Since you use SVN, it all depends actually on revision. I think I'd create a separate project for your third scenario and added a parameter where PM would provide revision number. Then based on that I'd tag sources etc. in my own task.
Regarding the other points, I think this is similar. Recently for web projects we started using MSDeploy, and in each stage build the MSDeploy package was created. Then there was a separate build called Deploy, that when forced allows us to select which package we want to deploy using MSDeploy.
Having several environments, however, started a little bit like overkill for managing with CCNet, and I'll be looking into kwakee at some time.