With EF code-based migrations, after performing a "Get Latest Version" of the code, it isn't always obvious if you have migrations to run. You could look in the migrations folder to see if there are any new ones, or you can just wait until something breaks and perhaps it will become obvious that your DB is out of date. Neither of those options are appealing to me.
I don't want to enable automatic migrations, and I also prefer not to have migrations always run on application startup because I may not be ready for them yet. I could run update-database every time I get latest (and don't have pending DB changes of my own), but that would be unnecessary most of the time.
Is there a simple way to know that my DB is out of date, particularly after pulling down the latest code?
Ideally I'm envisioning some sort of build event that only happens after a get latest.
You have to compare your model with the __MigrationsHistory table.
A simple way to do that is using update-databse with the -script switch. In that way you are doing the compare, but possible changes will not take effect instead they are generated as SQL in a new file.
update-database -script
Depending on your IDE and source control system I'm sure you can trigger a script from an "pull" or "get latest" event that does your compare.
Related
I have created database project in visual studio 2013 from existing database. Then I have done lot of changes in database project like modify stored procedures, post deployment script, table structure, etc . Now database project is ready to deploy. But I am worry if any script fails then How I can retain the original state though it build properly.
Please suggest that if any query fails then I want ROLLBACK the all changes that I have made in database project.
Firstly you need to trust your tools and either believe they will work or find other tools.
While you are building the trust I would add a create backup to the pre-deployment script or run a backup before you deploy then if anything goes wrong you can restore and figure out what went wrong.
As David said to roll-back, you would get the previously deployed dacpac and generate a new deployment script from that but fixing forward is almost always the correct thing to do rather than rolling back to a previous version.
ed
Have you been checking your changes into version control? If so, all you need to do is to revert back to the last known good version.
Or... simply work out why it's failing now and fix the root cause?
I used Db projects some time ago and as far as I remember the deploy script was wrapped in a transaction. It is possible to generate sql script without executing it. That setting was somewhere in DB project settings. You can take a look inside that script and make sure that it'll rollback on error.
Doing a backup would still be a recommended practice especially when you deploy to production.
when working on important scripts I developed a habit of always starting a transaction and commenting out the commit.
If you accidentally run it, it won't take effect. The commented out commit would only come out when the thing was done.
While this answer indicates that you CAN revert in source control (Assuming SSDT at this point) it would be nice to get a pointer to the exact process to do this. On a file by file basis the history works the same but how to revert the entire database at once isn't immediately obvious.
I'm working on creating a deploy script to migrate new development from our dev server to our uat server. Unfortunately, the devs that made the changes didn't script them out at they coded. The easiest way for me to approach this is to use SqlDeveloper's database diff functionality. It does a good job of highlighting the differences and creating a script that I can run on UAT. However, I've noticed that it doesn't take into account any dependencies. For example, a it will put a command to create a table below a table that references it in a foreign key constraint. Because the referenced table doesn't exist yet, the first table create command fails. I've seen this with views referencing packages, packages referencing packages, etc. Is there any easy to way either 1) force sql to export in a "smarter" order or 2) manually calculate the dependencies (ex: querying USER_DEPENDENCIES, etc) so that i can manually sort the file of create commands without resorting to trial and error? I guess we could consider purchasing a commercial product as long as it matched exactly what we are looking for.
Note: we will probably have to deploy to UAT multiple times in order to support testing by end-users. I am trying to automate this as much as possible so I don't have to manually recreate this script every single time!
Thanks!
EDITED: from original as frankly it was a poor question first time around....
We have a batch script called DEV.DBDeployment.DropCustomCreate.bat, as the name suggests this drops and creates our db from a fresh, a useful tool in Dev but we don't always want to drop the database, sometimes just get the latest changes.
It's worth noting currently every CI checkin triggers a build in TeamCity which pumps the current Major.Minor.BuildNumber.Revision (e.g. 1.0.123.1568) number in to all AssemblyInfo.cs files within all Visual Studio projects. This obviously allows us to stamp the resultant dll's with the build number, pretty standard stuff for sure. We also overwrite a BuildInfo.txt file in a similar way, most importantly this BuildInfo.txt file is included within every deployment package and sits within the RoundhousE\deployment folder and is referenced by /vf=%version.file% when we run rh.exe as mentioned above from the .bat file. So we're sorted for deploying to existing databases in Test and Prod.
However in dev the AssemblyVersion is always 0.0.0.0 in AssemblyInfo.cs, as is the version number in BuildInfo.txt, therefore how do devs stage their changes locally against their database. For example, with this setup when we run rh.exe all changes will be stamped with the version number 0.0.0.0. Is the expectation that in dev you will always drop and create? If that's the case I'm assuming we need TeamCity to checkin the BuildInfo.txt file so RoundhousE can reference it from source control when executed in dev?
Is there something I'm missing here?
I think we discussed this over here - https://github.com/chucknorris/roundhouse/issues/113
As you saying about the .bat file; that is a tool for roundhouse. You have to run that batch file again and again when you want to run your scripts. If you want to run scripts when you build the roundhouse database project then you have to configure that with certain steps. If you wish I can tell you if you replied.
Currently I'm using Visual Studio 2012 RC and SQL Server 2012 RTM.
I'd like to know how to re-deploy/re-create a test database for each test run.
Keep in mind I've a SQL Server database project for the database using Visual Studio 2012's template.
Actually I'm not very sure about an idea I got in my mind, but .testsettings file has Setup and cleanup scripts. Is this the way to go? For example, a PowerShell script reading the database project generated script and executing it against the database?
I guess there're better ways of doing that and it should be an out-of-the-box solution but I ignore it and Google doesn't help me in finding the right solution.
As mentioned you'll probably want to use the VS 2012 .Local.testsettings > Setup and Cleanup scripts to create / tear down you SQL Server database.
For the script you may want to use powershell with a .dacpac (rather than just a T-SQL script), since you are using a SSDT project. Here's a link to some example code - in particular you may want to take a look at the 'Deploy-Dac' command.
If you are unfamiliar with .dacpacs as the (build) output of SSDT-created database projects, take a look at this reference link.
Edit: Although this doesn't answer the question in a plain SQL Server way, an easy Entity Framework approach would be the following: I found that I could create and destroy my database every time correctly by using the DbContext.Database.CreateIfNotExists() and DbContext.Database.Delete() methods in my setup and cleanup phases of my tests.
The fastest solution, while a bit of a hack, is really straightforward. You can set the DB Projects properties under the debugging tab to "always re-created DB". Then test in two clicks, do a debug/build, then run all tests. You should get a freshly built DB on localDB for you tests to be ran against. You can also change the target for the debugging DB (again the DB projects properties) to whatever you want, so you can deploy to a .dacpac, or to an existing SQL DB or wherever. It means testing in two steps, and if your build is long, it may be annoying, but it works. Otherwise, I believe scripting is your only option.
I'm using the Database project in visual studio 2010 to generate a script to deploy my database (and it's changes). This works great.
Is there a way to have Visual Studio database project generate a rollback script as well as the deployment script.
I'm not looking for rolling back the transaction while deploying; but say I deploy it and my stored procedure has an overlooked performance issue that comes up a week later that requires a rollback to the previous version of the database.
Is there a way to generate the rollback script at build/deploy time that will undo whatever changes the deployment script made.
EDIT: If we ignore that I'm using a database project: What is a good way to have an upgrade and downgrade path for a database generated?
This generation needs to be part of an automated build process.
To create a rollback script While doing a schema compare using VS2010, It is as simple as swapping the db names specified in the source and target.
This way VS2010 would create a rollback script which would have drop statements against your stored proc.
I've not seen anything like that.
I think you need to reconsider this approach, as you'd still need to fix the stored proc in your database project, otherwise you'd just be re-deploying the "bad" version the next time you deploy. (I'm sure you're already aware of that, but it doesn't help to point out the obvious sometimes!)
If you need to restore an old version of the sproc to the server in the mean time, I would have thought that the easiest thing to do would be to get the previous version from source control and manually deploy that.
You could create a backup of the database before the release and then just restore from the backup if things go wrong. Obviously you'd also loose any data changes (either made as part of the release or subsequently) since the backup was taken.
Another idea I had was to create a snapshot before the release. The operation to create a snapshot is very light weight. I'm not sure you'd want to keep the snapshot for a week, but if the release went wrong then I think it would be quicker to restore from a snapshot than from a full backup. I would be interested to hear any comments people have on this idea.