What is the best practice to make changes in beans.xml file in production environment?
Lets imagine Hello World application with one Interface and 2 Implementations. After creating jar and running JVM how can we change implementation in beans.xml without opening jar and reassembling it afterwards?
Is it any trick that permits having beans.xml outside the jar, but without knowing the full file-system path?
On top of suggestion made by OrangeDog, I would suggest considering leaving the beans.xml inside your deployed application AND using the PropertyOverrideConfigurer mechanism.
Configure an override configurer to point to some fixed path outside of the application, something like /opt/configuration/something/override.properties would be ideal. If you can't use an absolute path, you can always try to play with file:../../.. style of paths. Then you can list only the changed values within the override.properties file.
Don't have the XML in a JAR. Your application should be deployed in at least a WAR, which the app server will probably "explode". The folder WEB-INF/classes will be added to the classpath. You can make changes to the exploded XML here and the app server should detect this and reload the app.
Edit:
If you are using plain Java SE, you can use a system property to pass the location of the XML file. System.getProperty("property") in code and -Dproperty=value on the command line.
Related
In our Linux environment applications are restarted periodically (the reasons aren't important here). It would be convenient for us to deploy new versions of an application by copying the application Spring Boot jar on top of the existing (old) jar thereby overwriting it and then simply wait for the application to restart (that is, the JVM running the application to restart).
However, this seems to not work. We get different kinds of errors - sometimes the app just hangs, sometimes we get a ClassNotFoundException. It's as if Spring Boot (or something inside Spring Boot) reopens the jar and expects it to be the same one it was when the application was originally started.
We had a look through Spring's common application properties, but didn't see anything appropriate. Is there a way to make this work? When we were using WAR files we configured the servlet container to unpack the WAR file and run from the unpacked version. Can we do something similar with Spring Boot?
First of all the errors you are experiencing can come from multiple sources. Usually replacing the file from a process is not a big problem as the whole file is loaded into memory before execution. Java is a little bit different, because the actual process that is running is the JVM and it only loads the jar file from disk. The JVM loads classes only on demand, this means if there was any class that was not loaded before it will try to load it and most likely fail, if the jar file is different. In the case of spring boot there are also other resources (such as HTML files) inside the jar file that are dynamically loaded.
You mentioned you are using a linux environment. If you can just replace your startup with a script you can just copy the jar and start it from the copied location:
#!/bin/bash
JAR_NAME="spring-boot.jar"
NEW_JAR_NAME=".$JAR_NAME" # Use an appropriate name here
cp $JAR_NAME $NEW_JAR_NAME
java -jar $NEW_JAR_NAME
rm $NEW_JAR_NAME
Now every time you start the application a copy is being made and started from there. You can replace the original jar and on the next restart the new application will load.
You coud also use rsync instead of cp to avoid copying the same jar twice, if the application is restarted multiple times without changing the jar.
It would be convenient for us to deploy new versions of an application
by copying the Spring Boot jar on top of the existing (old) jar and
then simply wait for the application to restart
Why would you do such a thing to yourself? You are trying solve a usecase that's against best practices, sound like asking for trouble just to avoid an app restart. When you are doing a deployment, you need to make sure the deployment went through, otherwise how will you troubleshoot if something goes wrong in your application, you will have one more variable in hand when you troubleshoot, i,e the uncertainty of current version of the code.
If you are having downtime while deploying (I am assuming thats why you want to limit the restarts), why don't you bring up another instance with the newer version of code and once its healthy shutdown the old one
Could be a silly question, but...
I have a Spring-based WAR application that runs 80% of the installations on Tomcat and the rest 20% on WebSphere.
I need to simply get the path of the folder where Spring's MultipartFilter (using Commons multipart resolver) stores files being uploaded. I have never set it manually, and it actually belongs to the Catalina work directory as I found out in my Tomcat installations.
For the moment, I just need to get that path. I have control of my application so no one is going to change it without notice.
I would like to know if there is a server-agnostic way to know where my Spring-based application is going to store Multipart files.
E.g. from this question I can see I can use catalina.base, but in Tomcat... not in JBoss or WebSphere. It could cost me a couple of if/else statements...
Spring tries to do it already when no default value is set
WebUtils.getTempDir(servletContext)
Which simply does:
return ((File)servletContext.getAttribute("javax.servlet.context.tempdir"));
So simply if no one overrides that location in the filter properties (and that is my case) I can rely on the default.
More in general, one may have to inspect the instance of DiskItemFileFactory to get the path to the repository
I'm working on a project that uses two maven projects (named core and webapp); core is built with JAR packaging and used for two different purposes: as a stand-alone app (essentially an executable JAR), and also embedded into webapp.
For its purpose as a stand-alone app, core needs to have its own logback configuration (a logback.xml file) that needs to be included on the classpath. Normal Maven convention would have me put it in src/main/resources/logback.xml. That works fine, but causes a problem when the core JAR is included in webapp. webapp needs to have its own logback configuration, but the container (tc Server or Jetty) is picking up the one from core.jar first.
I realize that logback can be told about a custom config location via a system property (-D on the command line) but that's not viable in a app container like Tomcat or Jetty.
I've read some other people asking about this situation, but none of the solutions I've seen sits well with me. One solution involved setting up a context listener that runs early in the webapp initialization and explicitly configures logback based on a <context-param>. That's a bit brutish in my opinion, and probably a hard sell to my fellow dev team when log4j "just works" in this situation.
I'm far from a Maven expert, so I'm hoping there is some elegant way to get Maven to help me here. Or perhaps some logback extension or add-on that makes it more web-app friendly. Or even a clever idea that I haven't thought of.
There are a number of possible solutions, but the easiest is to put the file in its own module and mark the dependency as provided. The, conspire to have it on the classpath when running the standalone version of the app.
The solution that we ended up using was to leave only the common "non-app" pieces (code and configuration) in core and then extract the other "app" pieces into a new module (batch-app).
The logging configuration only lives in the 2 app projects (webapp and batch-app) that depend on core. core has a logback-test.xml configuration in it, but that's excluded from the JAR that maven builds (since it's in the src/test/resources folder).
In my project I use Maven to package a web application to a WAR which is later deployed to Jetty using a custom Maven plugin on CentOS. The custom plugin is used by every project that deployes to a production environment. There is now a requirement that all static content (like web site text, properties) is packed outside of the WAR so that it can be changed in production without requiring a new release cycle. I am unsure how to achieve this externalization.
The Jetty server has the directory structure described in Jetty quick start guide. Currently, the web application already offers some .properties files which can be altered externally, and these reside in the resources/ directory. These files are moved here by the custom Maven plugin. The WAR resides in the webapp/ folder. One option for my problem is to use <packagingExcludes> of maven-war-plugin to not include e.g. *.xhtml and *.properties in the WAR. Later, I can use the custom Maven plugin to move excluded files to resources/ directory. But, I have a feeling this is not the correct way to externalize static content... Shouldn't xhtml files live in webapp/ folder while the only the properties file live in resources/ folder?
I have also researched the option of deploying the WAR as exploded, but I am unsure of the implications of such. Clearly, the changes in the exploded WAR files will be overwritten in the next deploy, but the idea is to do static changes both in development and production. Also, I am not sure how to achieve WAR "explosion", is it something that Jetty does for your WAR if configured in jetty.xml or do I have to extract the WAR before deploying?
Lastly, how do people serve static content in Jetty which can be altered in production? Do both the WAR and static files live side by side
The Jetty resources folder should not be used for application files. A J2EE web application (war) should be self-contained -- and in Jetty, reside only on the /webapps folder -- and its only binding to the container (Jetty servlet engine) is via the web.xml deployment descriptor.
Since property files may be read from the classpath and the Jetty resources folder is part of the system classpath, a property file there could be read by the web application class loader. Note that there are other ways to read property files as well and the Jetty resources folder should not be used for application properties. Also, the application may not be portable as other application servers have different forms of webapp classloader isolation.
If the below architecture approach does not work for you, then your only approach would be to expand (explode the war) in the /webapps folder and hope for the best when files are edited.
Tackling this from a different angle,
- if your web application depends on .properties and .xhtml files in order to function properly, then these files are probably not 'content'. The fact that there is a business process that requires them to to be updated ad hoc does not make them content.
- 'content' is something like text, images, and videos that is added, edited and removed by an administrative user. The application does not depend on it for correct execution, it merely reads and passes it on the browser.
Suggestions:
I would suggest that you deploy your application every time there is a change to the .xhtml or .properties files change. If the editors of these files are power business users, you might think of a git push-pull tool for them and a continuous build hook, so that when they make changes and push them to the git repository, the application gets tagged with a newer version and gets built and deployed. If there is a problem (tag not closed in xhtml), then it would be easy to roll back to the last tag.
Alternately, if the changes are minor (such as text descriptions), then modify the application to read them from an arbitrary external file (outside the webapp) -- a location that is provided to the webapp on startup. You can then package a 'default' version of the file in the webapp, but the code would attempt to look in the specified external location first.
I discovered that you can add HTML tags to properties and later use <h:outputFormat> to fetch these properties with parameters. Also, you can do pretty neat stuff with property files as described in MessageFormat API.
My solution was to keep the .xhtml files inside the WAR, but use simple HTML snippets properties from the default resource bundle which is based on a .properties file. These properties were included in the .xhtml using <h:outputFormat>and <h:outputText>. This allows the user to add simple styling like bold and underline to the snippets.
The properties file are copied to the Jetty resource folder using the custom Maven plugin, so I have kept the .properties files in the WAR. For some reason the Jetty resource folder has precedence over the packed .properties files, so this works out fine. Also, as Akber pointed out, I will have the default versions of the properties available if for some reason the WAR was moved to some other application server where the resource folder is not available.
Of course, with this approach the code can break if malformed HTML is placed inside the snippet properties, as pointed out by Akber, but it works for our application as it is very small. I may never have done this if this was a much larger application, but then I might have gone for a database based solution for adding static text (like Joomla/Drupal/Wordpress).
We have multiple projects, and there are spring configuration XMLs in all of them. We're contemplating 2 options:
A) Loading the spring files from the classpath, by using import "classpath:a.xml". In this option our springjunit4classrunner and our main console application both happily find the files. However, when you need to change or inspect some XML, it's well hidden in some obscure jar.
B) Creating a "conf" folder, and by using some maven magic gathering all the XML files from all the projects there. This is good for the main application, but the unittests don't like this setting, as obviously we don't build the "conf" folder before running unittests, and they fail to find the dependent projects XMLS...
(This is a spring re-take on Is it bad practice to include properties/configuaration files within jars?)
In the past, I've gone with <import resource="classpath*:beans.xml"/> or similar, i.e. import every beans.xml file the classloader can find in the root of the classpath. In turn, those beans.xml files import their own specific config files as required.
It requires discipline, so that you don't get big spider webs of Spring configs, but if you keep it simple, then it makes integration very easy, so that you don't need to maintain complex lists of config file imports, and the build process stays simple also.
It's interesting that you mention that prior question, because I don't consider Spring beans files to be "configuration". Rather, they're part of the application - they cannot and should not be separated from the Java code, because they're intimately coupled with it.
If those beans files require genuine configuration, e.g. from a properties file, then yes, those probably should be kept separate.