How to deal with database initialization? - installation

As also described here, I'm trying to determine the best way to initialize and update my application's database. I use EclipseLink-JPA2. I distribute a NetBeans platform application.
Considered options:
use create-tables ddl-generation:
The problem with this is that everytime the application runs it will throw exceptions, failing to create the tables. It will be useful only at setup time. This would be similar to placing checking code in the module restored() method.
include the database with the application distribution: the ddl-generation strategy becomes do nothing. I could still use the JPA (at development time) to generate the database files (embedded Java DB).
The best solution would be for the installer/setup (first-time) to call initializing code that creates the database. This precisely what I do with JWS in here. But I don't know how to do that without JWS. A script/jar executed by the installer?

Distribute your NetBeans Platform application via JWS.
It seems like you already have this problem sorted out with JWS, so just use that solution to deploy your NetBeans platform app.
EDIT
An alternative approach would be to use a Module Installer that checks for the existence of a config file. If the file exists then the application has been run before and the setup is not required. If the file doesn't exist then run your setup steps.
Have a look at this tutorial, they're using a module installer to create a login dialog.

Related

Node (maven) to deploy the application to several environments

On Jelastic, I created a node for building an application (maven), there are several identical environments (NGINX + Spring Boot), the difference is in binding to its database and configured SSL.
The task is to ensure that after building the application (* .jar), deploy at the same time go to these several environments, how to implement it?
When editing a project, it is possible to specify only one environment, multi-selection is not provided.
it`s allowed to specify just one environment
We suggest creating a few environments using one Repository branch, and run updates by API https://docs.jelastic.com/api/#!/api/environment.Vcs-method-Update pushing whole code to VCS.
It's possible to use CloudScripting technology for attaching custom logic to onAfterBuildProject event and deploying the project to additional environments after build is complete. Please check this JPS as an example of the code syntax. Most likely you will need to use DeployProject API method.

Websphere application modify web.xml doesn't work

I have deployed applications in Websphere 8.5, and I want to modify web.xml, but it seems not working. What I am supposed to do?
While there are documented ways of updating enterprise application files, those were conceived for multi-server deployments and partly are legacy of the previous decade of monster application servers. They are inconvenient for making changes to local development server and it wastes a lot of time.
Upon application deployment, WAS creates (copies/updates) deployment descriptors in the config directory. Then web.xml is used from that location.
You are probably changing web.xml in the location where original application files are kept, thus no effect. You should change the one in config\cells\<cell_name>\applications\<ear_name>\deployments\<app_name>\<war_name>\WEB-INF.
WebSphere writes a second file named web_merged.xml. If you only update the web.xml and replace it you will not update the runtime file used by the container.
Best result I had was using the single file upload function provided within the Admin console or to use the wsadmin or jacl cmd.
The proper way is to update application via console/script.
But I'm assuming you are editing file directly (very hard to guess from your description, I've asked you to describe your procedure).
You need to restart the application to pick up changes in web.xml. See the Hot deploy in WAR files
Two ways:
Update web.xml with WebSphere Web Console
Update web_merged.xml at the same time, and also update the two files in config\cells\<cell_name>\applications\<ear_name>\deployments\<app_name>\<war_name>\WEB-INF

Per-test sample data generation for load testing?

I have a set of load tests, each testing a specific section of Sharepoint site. How can I automate the generation of sample data prior to each test?
I know that load tests allow me to specify tests that are run before/after a specific virtual user's test mix, but those are meant for user-specific setup like logging on. I want to create sample data for all users.
I can put my setup code into the Team Build template, but that doesn't take into account whether the test using the sample data is in the test list being run or not. If I modify the test list, I would also need to modify the build template as well. I cannot put the setup code into a test that is run before the load test either, because load tests must be run with the .Net 4.0 framework while accessing Sharepoint's API requires 3.5.
I could rewrite my sample data generation logic with Powershell and start it from a .Net 4.0 unit test. Is it my best option or is there a more elegant way?
I think you question is more TeamBuild related than SharePoint. For short, the best way how to setup (and teardown) testdata in SharePoint is using PoSh in conjunction with the Microsoft.PowerShell.SharePoint SnapIn.
You can start your PowerShell scripts from Everywhere in TeamBuild. That's actually how I extend the default TeamBuild template. I've create a single custom build template, this template contains numerous PoSh hooks. While setting up a new project you can easily hook your scripts everywhere you need.
To be really independent from any .NET FX version you could create your sampledata using REST DataServices (ListData.svc)

A good strategy for knowing the versions of deployed applications?

In order to know which revision number the application is built from, we use to give the ears we deploy to Glassfish names like myapp_2012-01-20_rev22123.ear. Then we can simply login to Glassfish and see what version is deployed in the web interface (as the appname is the name of the ear file). A downside of this approach is that we need to do a manual undeploy/redeploy to update the name...
But I would like to script the undeploy/deploy process, and having each version of an ear get a different name is not very suitable to scripting this redeployment process. Glassfish 2 does not support the "list applications" goal that Glassfish 3 has, which I could have used to retrieve the application name to undeploy.
So is there any good strategy that will easily allow us to see what version is deployed of an application, and that does not suffer from the above fault?
It would be preferable if this meant we did not have to change the existing applications (like add a jsp page or something to show the current scm revision), but a change in a Maven build script would be acceptable.
I faced a similar issue, I finally came around it by using maven-buildnumber-plugin and writing a simple servlet to get build information. You can find the details in the blog post I made.
Why not use the built-in GlassFish Server versioning to assign a version number at deploy time? This will also enable you to rollback to prior versions. For example:
asadmin deploy --name MyApplication:2012-01-20_rev22123 MyApplication.ear
There is more information on application versioning here:
http://docs.oracle.com/cd/E18930_01/html/821-2417/gihzx.html#gkhhv
Hope this helps.

how do I use web deploy to deploy multiple sites on localhost with different ports

I have a solution that contains multiple integration test projects and one web application project. each integration project connects to the web application when running the tests. I would like for each test project to access the website with its own database connection. I have been trying to use the web deploy functionality built into visual studio. However I have been unable to figure out what I need to add to either the deployment package that is created and/or the post build event for the test projects to declare the binding port for the website when deployed. For example, I want integration project A to create and access the website located at http://localhost:83 and integration project B to create and access the website located at http://localhost:82. Could someone please explain:
Is there anything I need to do the deployment package ?
What do I need to add to my post-build events for my integration projects when deploying the package, so that the website is created at the correct port when building the project?
Update:
I'm wanting to deploy the same site to two different locations on my machine so that I can run both sets of integration tests at the same time.
Update 2:
I have researched the web deploy tool and it allows you to specify parameters that modify what is deployed when you call it from the command line. However I have found the documentation very confusing. http://technet.microsoft.com/en-us/library/dd568968(WS.10).aspx
Update 3:
I expect these to be two different websites, each pointing to there own database. If possible I would like a single package that can be deployed using msdeploy. Which will then be called in a post build event from each of the integration test projects. I would like to specify the connection string and deployment location from the post build script of the integration project.
you can try with webdev.server included in visual studio. VisualStudio use this for start a webserver when you debug. With this you can start a webserver in the desire port (if the port is not used currently).
I made a bat file for change some options.
check it.
::Begin of bat file
cd C:\Program Files\Common Files\microsoft shared\DevServer\10.0\
WebDev.WebServer40.exe /port:80 /path:"C:\PATHTOYOURWEBPROJECT" /vpath:"/NAMEOFYOURWEBPROJECT"
::End of bat file
You can acces in: http://localhost:80
I use the webserver40, but if you don't have net.4 or vs2010 you can try to find webserver[ xx version].exe
I hope that this will be helpful, and sorry for my broken english.
First off, you're approaching this the wrong way.
> I would like for each test project to access the website with its
own database connection.
Who is creating the DB connection? Your web site or the test project? For rest of your question to make sense, I presume its the web site (otherwise, Project A and Project B cannot share a connection out of the box).
If your website is making the connection, unless you're caching or having a static connection, there will be a new connection made as each request runs your your site on a new thread. Another simpler alternative is to take a query param and initiate a new connection based on that. If you seed it off the caller, you can also use it for more detailed logging.
Web Deployment projects are meant for deploying to integration servers, so that means you cannot access them by http://localhost... but the full FQDN of the server.
Most importantly, http://localhost:82/myApp and http://localhost:83/myApp are two different sites (unless you redirect from one of them to another one which in itself can cause additional issues) running the same codebase.
Having said that, you would then need to deploy your website twice and then all you need is to change the config/settings entry in Project A and B to point to these to different sites.
Hope this makes sense.
You can define virtual host configuration.
Refer this guide for more information.
http://docs.jboss.org/jbossas/guides/webguide/r2/en/html/ch07.html

Resources