We are trying to install a CI Platform with (Jenkins,sonar,eclipse ...).
So that every developer can make analysis on his code before commit, I'm wondering between two alternatives :
running local analysis with the sonar plugin.
install the different plugins that sonar use (findbug,pmd,checkstyle ...) and configure them to meet the sonar configuration.
I'm not sure which alternative to use? I used to work with findbugs,pmd, checkstyle in eclipse and they look great.
Can you tell me which is the best alternative?
Thanks in advance.
Regards.
With Sonar plugin you can manage the violations like:
Create a review
Mark a violations as false positive or fixed
View the hot classes and hot violations
View yours reviews
If you use separate plugins you have to go sonar web to do that.
The great advance of sonar is the reviews.
Other question is how many projects you have and will have. I currently work with more than 70 projects and many profiles. Is more simple to me run analysis with one plugin, because I need just add the server and find the project. With other plugin you need add the link for each project in each plugin configuration.
Why not install the Sonar Eclipse plugin?
This was designed to solve the following problems:
Sonar does not support parallel analysis of the same project. This issue rules out the option of each developer running Sonar locally. (See SONAR-2761, SONAR-3306)
You don't really want developers uploading metrics and source code into the Sonar database. They could be working on an uncommitted workspace and would therefore cause both inaccuracies and confusion if Sonar is being used for code review.
Sonar is really designed to be run from a continuous integration server (like Jenkins), building code that has been submitted onto a shared codestream (or branch)
The big advantages of using the Eclipse plugin are:
True local analysis, no updates of the Sonar database
Configuration of the other tools is retrieved from the Sonar server and jars automatically downloaded.
Centralized management of Sonar quality profiles
Related
I recently started using SonarQube on my Maven Java Project. The problem is that I modified a single line in a Class which causes Sonar failing to load the project into the database. Why does Sonar know that my project is checked into a SVN and why does mvn sonar:sonar fails with "can not blame XXX on line YYY" if a class is not committed into the SVN?
I am the only one working on this and I dont want to check in every single change or experimental code snipet. How can I turn this feature off?
Thank you!
SCM integration (which detects that your projects uses SVN) allows SonarQube to track changes to source code, e.g to compute coverage on new code, new issues etc.
A standard analysis is supposed to be performed from commited code (e.g by a continuous integration server) to serve as a reference.
In your case, I would advise you to use the incremental mode: this will allow you to perform an analysis on your local code changes, which will not be persisted to the central server, but which will show you which issues your new code introduced/fixed (this is the analysis mode used by the IDE integration plugins).
I have a simple problem, with a simple answer probably, but I can't find what is it. We want to deploy SonarQube along with Checkstyle and some other tools, but we can't find out is it meant for a centralized, server deployment, or on each developer machine? All tutorials show installations on separate machines and being used in the localhost, while there is a public instance example, and the requirements and specs certainly look service-like.
On the other hand, I'm not getting how do the developers submit their code for checks if it is on a server.
So, in short, how is it deployed? Any checklist or something similar would be of great help.
The SonarQube "runtime" architecture has several elements:
SonarQube server. It contains a database (e.g., MySql) and an
embedded web server (Tomcat). The SonarQube server stores the
results of analyses (the metrics), but does not execute the code
analyses. This server provides a web UI that shows the dashboard of
the projects, various metrics and drill down into code, admin options. It uses a pluggable architecture--you can add/remove funcitionality via plug-ins.
Program that runs code analysis on the developer machine. There are options: (a) if they are using Eclipse or IntelliJ, they can use the respective SonarLint plug-in, which provides configuration properties, menu options to run analysis, a view to show violations, etc.; (b) developers can also run code analysis via maven (mvn sonar:sonar) or gradle (gradlew sonarqube); (c) developers can execute the various code analyses through a program called SonarQube Runner.
All these options of programs that run the analysis on the developer machine need to be configured to communicate with a SonarQube server. For example, when you run code analysis in IntelliJ using SonarLint, the metrics will be uploaded to the server. This server is typically shared by all developers, but it can also be localhost.
Program that runs code analysis on the CI/CD server. The job/pipeline that builds a software project can be configured to run SonarQube code analysis. It can be done via maven or gradle just like on the developer's machine, or via a plug-in. There are SonarQube CI plug-ins for Jenkins, Hudson, Bamboo, and others. Depending on the size of your project, you may want to configure the code analysis to run once a day only, and not upon each code commit or changes to dependencies. The SonarQube code analysis executed on the CI server will likewise send the generated metrics to the SonarQube server.
The SonarQube architecture documentation is very poor (not to say absent), so it's hard to get the big picture. I hope this helps.
SonarQube (formerly just "Sonar") is a server-based system. Of course you can install it on your local machine (the hardware requirements are minimal). But it is a central server with a database.
Analyses are performed by some Sonar "client" software, which could be the sonar runner, the sonar ant task, the sonar Eclipse plugin etc. The analysis results can be automatically uploaded to the server, where they can be accessed via the sonar Web application.
In an environment with many developers, you should run a build server (e.g. Hudson or Jenkins), which performs automatic sonar analyses as part of the nightly build. Other schedules are possible, but the developers should know when they can expect updates of the server-side analysis results. The results of the automated analysis can be displayed in the individual developer's Eclipse editor by way of the sonar Eclipse plugin.
The architectural documentation on Sonar is quite sparse. I've looked for a picture to visualize what I just described, but could not find one ...
We have upgarded sonar server 3.5.1. In order to use sonar I have upgraded my local eclipse to Juno(4.2) and sonar plugin to 3.1. Sonar server is able to exclude some packages usign sonar.excustions key and does not show violations for those packages. But when I run report for sonar local analysis I see even though I added sonar.exclusions from local properties it does show violations for exluded files. How i can resolve this issue.
Thanks in advance.
Sapana.
There's currently a limitation on Sonar side (see http://jira.codehaus.org/browse/SONAR-4265) that prevents modules to retrieve all their related settings when analyzed independently from their root project.
Feel free to watch and vote for the ticket.
I have configured one project in sonar and integrated sonar with maven for build time analysis of the project.
After analysis, report is generated and uploaded to Sonar for browsing. But once another user compiles the same project their report overwrites mine.
Basically I want that one user's report on one project is not overwritten by report from other user. A user must be able to see their current violations independently. Is it possible in Sonar?
Sonar stores it's analysis on a daily basis, which explains why it's kind of pointless to run analysis several times in a day. Each analysis run will overwrite that day's existing results, which in turn spoils ongoing statistical analysis.
I would suggest running Sonar, from a dedicated build server like Jenkins (which has a Sonar plug-in). This daily analysis will populate the Sonar database and keep the project dashboard current. This architecture also enables you to keep the database credentials confidential.
Obviously developers would like to see the results of their bug fixing. For that I'd recommend running the Sonar Eclipse plug-in. The latest version will run the same Sonar analysis locally. Recent versions of Sonar also enable you to assign violations to developers for resolution.
This is not possible, last performed analysis will always be the one you browse in the interface. However, I guess what you need is the Issue Report Plugin which will enable analysis to store results locally, with the dry-run option.
This way your developers will be able to run an analysis on their code and see the violation delta without pushing the results.
You can do it by explicitly setting the below properties in pom.xml
<properties>
<sonar.projectKey>Test</sonar.projectKey>
<sonar.projectName>Test</sonar.projectName>
</properties>
Every user should set different projectKey and projectName, if you want that one user's report on one project is not overwritten by report from other user.
We are working on a web project from scratch and are looking at the following static code analysis tools.
Conventions (Checkstyle)
Bad practices (PMD)
Potential bugs (FindBugs)
The project is built on Maven. Instead of using multiple tools for the purpose, I was looking at a single flexible solution and came across SonarQube.
Is it true that we can achieve the results from Checkstyle, PMD and Findbugs with SonarQube?
Sonar will run CheckStyle, FindBugs and PMD, as well as a few other "plugins" such as Cobertura (code coverage) by default for Java projects. The main added value, however, is that it stores the history in a database. You can then see the trend. Are you improving the code base or are you doing the opposite? Only a tool with memory can tell you that.
You should run Sonar in your CI system so that even things that take some time to execute (such as CPD – copy paste detector) can run. And you'll have your history. Whereas with an Eclipse plugin, for example, you'll detect violations sooner – which is great – but you will be tempted to run it less often if it starts taking too long, or run less "quality plugins" (such as skipping CPD or skipping code coverage analysis). And you won't have history.
Also, Sonar generates visual reports, "Dashboard" style. Which makes it very easy to grasp. With Sonar in Jenkins, you'll be able to show developers and your management the effects of the work that was performed on the quality of the code base over the last few weeks and months.
Sonar uses these 3 tools as plugins and aggregates the data from all three giving addition value by showing graphs and such from these tools. So they are complementary to sonar.
Yes and no. In addition to the other answers.
SonarQube is currently on the way to deprecate PMD, Checkstyle and Findbugs and use their own technology to analyze Java code (called SonarJava). They do it, because they don't want to spend their time fixing, upgrading (or waiting on it) those libraries (e.g. for Java 8), which for example uses outdated libraries.
They also got a new set of plugins for your personal IDE called SonarLint.
Sonar is great, but if you want to use the mentioned tools separately and still have nice graphs, you can use the Analysis Collector Plugin as part of your Jenkins CI build. A slight advantage of this is that you can check in your PMD/Findbugs/Checkstyle configuration into your SCM and have it integrated into your Maven build, rather than relying on a separate Sonar server.
... a few years later: no, it is not! SonarQube supposes to be able to cover all the rules with its own analyzer, but there are still rules from PMD or CheckStyle not covered by SonarQube. See for example: PMD ReturnFromFinallyBlock.
Sonar is much more than these tools alone.
The greatest benefits is the gui, which lets you configure anything easily.
The statistics it offers are very detailed (lines of code etc).
And it even offers great support for test coverage etc :)
Here you can take a good look:
http://nemo.sonarsource.org/
I would still use these tools in addition to sonar because they can fail the maven build when someone violates a rule. Where as sonar is more retrospective.
Well at least since SonarQube 6.3+ it seems to be that Findbugs is (at the moment) no longer supported as a plugin. Sonarsource is working on replacements of Findbugs-rules with its own Java-plugin.
They even had a list for the replacement status of each rule here, but it got removed by now.
See https://community.sonarsource.com/t/where-is-dist-sonarsource-com-content/5353 for more details.