Copyright (c) 2006, 2017 Oracle and/or its affiliates. All rights reserved.

1. Introduction
2. Requirements
3. Usage
4. OpenGrok install
5. OpenGrok setup
6. Optional Command Line Interface Usage
7. Change web application properties or name
8. Information for developers
9. Tuning OpenGrok for large code bases
10. Authors
11. Contact us

OpenGrok is a fast and usable source code search and cross reference engine, written in Java. It helps you search, cross-reference and navigate your source tree. It can understand various program file formats and version control histories of many source code management systems.

Official page of the project is on: http://opengrok.github.com/OpenGrok/

• Latest Java (at least 1.8) http://www.oracle.com/technetwork/java/
• A servlet container like Tomcat (8.x or later) supporting Servlet 2.5 and JSP 2.1 http://tomcat.apache.org/
• Exuberant Ctags or Universal Ctags http://ctags.sourceforge.net/ https://ctags.io/
• Source Code Management installation depending on type of repositories indexed
• If you want to build OpenGrok:
  • Ant (1.9.4 and later) http://ant.apache.org/
  • JFlex http://www.jflex.de/
  • Netbeans (optional, at least 8.2, will need Ant 1.9.4) http://netbeans.org/

OpenGrok usually runs in servlet container (e.g. Tomcat).

SRC_ROOT environment variable refers to the directory containing your source tree. OpenGrok analyzes the source tree and builds a search index along with cross-referenced hypertext versions of the source files. These generated data files will be stored in directory referred to with environment variable called DATA_ROOT.

OpenGrok has a concept of Projects - one project is one directory underneath SRC_ROOT directory which usually contains a checkout of a project sources. (this can be branch, version, ...)

Projects effectively replace the need to have more web applications, each with opengrok .war file. Instead it leaves you with one indexer and one web application serving multiple source code repositories - projects. Then you have a simple update script and simple index refresher script in place, which simplifies management of more repositories.

A nice concept is to have a naming convention for directories underneath SRC_ROOT, thereby creating a good overview of projects (e.g. name-version-branch).

For example, the SRC_ROOT directory can contain the following directories:

• openssl-head
• openssl-0.9.8-stable
• openssl-1.0.0-stable

Each of these directories was created with cvs checkout command (with appropriate arguments to get given branch) and will be treated by OpenGrok as a project.

Deployed OpenGrok can receive couple of messages through the active socket which usually listens for the main configuration file. These are used in the web application and displayed to the users. One can easily notify users about some important events, for example that the reindex is being in progress and that the searched information can be inconsistent.

The OpenGrok comes with a tool which allows you to send these messages without any problem. It is called Messages and it is located under the tools directory. See the file for usage and more information.

Any message can use tags which makes it more specific for the application. Messages which tag match some OpenGrok project are considered project specific and the information contained in them are displayed only for the specific projects.

There is a key tag main which is exclusive for displaying messages on the OpenGrok landing page - like a common information.

Currently supported message types:

1. NormalMessage (normal) – this message is designed to display some information in the web application. Use tags to target a specific project.

2. AbortMessage (abort) – this message can delete some already published information in the web application. Use tags to restrict the deletion only to specific projects.

3. StatsMessage (stats) – this message is designed to retrieve some information from the web application.

   The purpose of the message is specified in the text field as one of:

   • reload – the application reloads the statistics file and returns the loaded statistics
   • clean – the application cleans its current statistics and returns the empty statistics
   • get – the application returns current statistics

4. ConfigMessage (config) – this message performs some configuration communication with the webapp, depending on tag.

   • setconf – tag sends config to webapp and requires a file as an argument.
   • getconf – tag retrieves the configuration from the webapp.
   • set – tag sets particular configuration option in the webapp.
   • auth – tag requires "reload" text and reloads all authorization plugins.

5. RefreshMesssage (refresh) – sent at the end of partial reindex to trigger refresh of SearcherManagers.

6. ProjectMessage – used for adding/deleting projects and partial (per-project) reindex.

• add – adds project(s) and its repositories to the configuration. If the project already exists, refresh list of its repositories.
• delete – removes project(s) and its repositores from the configuration. Also deletes its data under data root (but not the source code).
• indexed – mark the project(s) as indexed so it becomes visible in the UI
• get-repos – get list of repositories in the form of relative paths to source root for given project(s)
• get-repos-type – get repository type(s) for given project(s)

6. RepositoryMessage – used for getting repository info.

• get-repo-type – get repository type

The file <package_name>.p5p you can easily use as a new publisher for the pkg command.

pkg install --no-refresh -g /path/to/file <package_name>.p5p opengrok

You can also update OpenGrok software with the *.p5p file by running a command

pkg update --no-refresh -g /path/to/file/<package_name>.p5p 'pkg://opengrok/*'

To setup OpenGrok it is needed to prepare the source code, let OpenGrok index it and start the web application.

Source base should be available locally for OpenGrok to work efficiently. No changes are required to your source tree. If the code is under source control management (SCM) OpenGrok requires the checked out source tree under SRC_ROOT.

By itself OpenGrok does not perform the setup of the source code repositories or synchronization of the source code with its origin. This needs to be done by the user or by using automatic scripts.

It is possible for SCM systems which are not distributed (Subversion, CVS) to use a remote repository but this is not recommended due to the performance penalty. Special option when running the OpenGrok indexer is needed to enable remote repository support (-r on).

In order for history indexing to work for any SCM system it is necessary to have environment for given SCM systems installed and in a path accessible by OpenGrok.

Note that OpenGrok ignores symbolic links.

If you want to skip indexing the history of a particular directory (and all of it's subdirectories), you can touch .opengrok_skip_history file at the root of that directory. If you want to disable history generation for all repositories globally, then set OPENGROK_GENERATE_HISTORY environment variable to off during indexing.

For *nix systems there is a shell script called OpenGrok which simplifies most of the tasks. It has been tested on Solaris and Linux distributions.

First please change to opengrok directory where the OpenGrok shell script is stored (can vary on your system).

Note that now you might need to change to user which owns the target directories for data, e.g. on Solaris you'd do:

pfexec su - webservd
cd /usr/opengrok/bin

and run

./OpenGrok deploy

This command will do some sanity checks and will deploy the source.war in its directory to one of detected web application containers. Please follow the error message it provides.

If it fails to discover your container, please refer to optional steps on changing web application properties below, which explains how to do this.

Note that OpenGrok script expects the directory /var/opengrok to be available to user running opengrok with all permissions. In root user case it will create all the directories needed, otherwise you have to manually create the directory and grant all permissions to the user used.

During this process the indexer will generate OpenGrok XML configuration file configuration.xml and sends the updated configuration to your web app.

The indexing can take a lot of time. After this is done, indexer automatically attempts to upload newly generated configuration to the web application. Most probably you will not be able to use Opengrok before this is done for the first time.

Please change to opengrok directory (can vary on your system)

cd /usr/opengrok/bin

and run, if your SRC_ROOT is prepared under /var/opengrok/src

./OpenGrok index

otherwise (if SRC_ROOT is in different directory) run:

./OpenGrok index <absolute_path_to_your_SRC_ROOT>

The above command attempts to upload the latest index status reflected into configuration.xml to a running source web application. Once above command finishes without errors (e.g. SEVERE: Failed to send configuration to localhost:2424), you should be able to enjoy your OpenGrok and search your sources using latest indexes and setup.

It is assumed that any SCM commands are reachable in one of the components of the PATH environment variable (e.g. git command for Git repositories). Likewise, this should be maintained in the environment of the user which runs the web server instance.

Congratulations, you should now be able to point your browser to http://<YOUR_WEBAPP_SERVER>:<WEBAPPSRV_PORT>/source to work with your fresh OpenGrok installation! :-)

At this time we'd like to point out some customization to OpenGrok script for advanced users. A common case would be, that you want the data in some other directory than /var/opengrok. This can be easily achieved by using environment variable OPENGROK_INSTANCE_BASE.

E.g. if OpenGrok data directory is /tank/opengrok and source root is in /tank/source then to get more verbosity run the indexer as:

OPENGROK_VERBOSE=true OPENGROK_INSTANCE_BASE=/tank/opengrok
./OpenGrok index /tank/source

Since above will also change default location of config file, beforehand(or restart your web container after creating this symlink) I suggest doing below for our case of having OpenGrok instance in /tank/opengrok:

ln -s /tank/opengrok/etc/configuration.xml /var/opengrok/etc/configuration.xml

More customizations can be found inside the script, you just need to have a look at it, eventually create a configuration out of it and use OPENGROK_CONFIGURATION environment variable to point to it. Obviously such setups can be used for nightly cron job updates of index or other automated purposes.

There is inherent time window between after the source code is updated (highlighted in step 5.1 above) and before indexer completes. During this time window the index does not match the source code. To alleviate this limitation, one can kick off update of all source repositories in parallel and once all the mirroring completes perform complete reindex. This does not really help in case when some of the source code repositories are slow to sync, e.g. because the latency to their origin is significant, because the overall mirroring process has to wait for all the projects to finish syncing before running the indexer. To overcome this limitation, the index of each project can be created just after the mirroring of this project finishes.

Indexing of all projects (i.e. OpenGrok index /var/opengrok/src) in order to discover new and remove deleted projects from the configuration can be avoided. One can start with no index and no projects and then incrementally add them and index them.

It basically works like this:

1. bootstrap initial configuration: OpenGrok bootstrap

• this will create /var/opengrok/etc/configuration.xml with basic set of properties. If more is needed use:

  Messages -n config -t set "set propertyXY = valueFOO"
  

2. add a new project foo:

Messages -t foo -n project add

• the project foo is now visible in the configuration however is not yet searchable. Store the config in a file so that indexer can see the project and its repositories:

  Messages -n config -t getconf > /var/opengrok/etc/configuration.xml
  

3. index the project. It will become searchable after that.

OPENGROK_READ_XML_CONFIGURATION=/var/opengrok/etc/configuration.xml
OpenGrok indexpart /foo

4. make the project indexed status of the project persistent so that if webapp is redeployed the project will be still visible:

Messages -n config -t getconf > /var/opengrok/etc/configuration.xml

If an add message is sent that matches existing project, the list of repositories for that project will be refreshed. This is handy when repositories are added/deleted.

When running the indexer the logs are being written to single file. Since multiple indexers are being run in parallel in step 2, their logs have to be separated. To do this, create logging.properties file for each project using the /var/opengrok/logging.properties file as template. The only line which can differ would be this:

java.util.logging.FileHandler.pattern = /var/opengrok/log/myproj/opengrok%g.%u.log

Note the path component myproj which separates the logs for given project to this directory. The creation of the per-project directory and the logging.properties file can be easily done in a script.

The command used in step 2 can thus look like this:

OPENGROK_LOGGER_CONFIG_PATH=/var/opengrok/myproj.logging
OPENGROK_READ_XML_CONFIGURATION=/var/opengrok/etc/configuration.xml
OpenGrok indexpart /myproj

The last argument is path relative to SRC_ROOT.

If you installed OpenGrok from the OSOLopengrok package, it will work out of the box. Should you need to configure it (e.g. because of non-default SRC_ROOT or DATA_ROOT paths) it is done via the opengrok property group of the service like this:

svccfg -s opengrok setprop opengrok/srcdir="/absolute/path/to/your/sourcetree"
svccfg -s opengrok setprop opengrok/maxmemory="2048"

Then make the service start the indexing, at this point it would be nice if the web application is already running.

Now enable the service:

svcadm enable -rs opengrok

Note that this will enable tomcat service as dependency.

When the service starts indexing for first time, it's already enabled and depending on tomcat, so at this point the web application should be already running.

Note that indexing is not done when the opengrok service is disabled.

To rebuild the index later (e.g. after source code changed) just run:

svcadm refresh opengrok

The service makes it possible to supply part of the configuration via the opengrok/readonly_config service property which is set to /etc/opengrok/readonly_configuration.xml by default.

Note: before removing the package please disable the service. If you don't do it, it will not be removed automatically. In such case please remove it manually.

There are 2 (or 3) steps needed for this task.

• Option 1. OpenGrok: There is a sample shell script OpenGrok that is suitable for using in a cron job to run regularly. Modify the variables in the script to point appropriate directories, or as the code suggests factor your local configuration into a separate file and simplify future upgrades.

• Option 2. opengrok.jar: You can also directly use the Java application. If the sources are all located in a directory SRC_ROOT and the data and hypertext files generated by OpenGrok are to be stored in DATA_ROOT, run

  java -jar opengrok.jar -s $SRC_ROOT -d $DATA_ROOT

  See opengrok.jar manual below for more details.

To configure the webapp source.war, look into the parameters defined in web.xml of source.war file and change them (see note1) appropriately.

• HEADER – the fragment of HTML that will be used to display title or logo of your project

• SRC_ROOT – absolute path name of the root directory of your source tree

• DATA_ROOT – absolute path of the directory where OpenGrok data files are stored

  • File header_include can be created under DATA_ROOT. The contents of this file will be appended to the header of each web page after the OpenGrok logo element.
  • File footer_include can be created under DATA_ROOT. The contents of this file will be appended to the footer of each web page after the information about last index update.
  • The file body_include can be created under DATA_ROOT. The contents of this file will be inserted above the footer of the web application's "Home" page.
  • The file error_forbidden_include can be created under DATA_ROOT. The contents of this file will be displayed as the error page when the user is forbidden to see a particular project with HTTP 403 code.

OpenGrok can use path descriptions in various places (e.g. while showing directory listings or search results). Example descriptions are in paths.tsv file (delivered as /usr/opengrok/doc/paths.tsv by OpenGrok package on Solaris).

The paths.tsv file is read by OpenGrok indexing script from the configuration directory (the same where configuration.xml is located) which will create file dtags.eftar in the index subdirectory under DATA_ROOT directory which will then be used by the webapp to display the descriptions.

The file contains descriptions for directories one per line. Path to the directory and its description are separated by tab. The path to the directory is absolute path under the SRC_ROOT directory.

For example, if the SRC_ROOT directory contains the following directories:

• foo
• bar
• bar/blah
• random
• random/code

then the paths.tsv file contents can look like this:

/foo  source code for foo
/bar  source code for bar
/bar/blah source code for blah

Note that only some paths can have a description.

web.xml is the deployment descriptor for the web application. It is in a Jar file named source.war, you can change it as follows:

• Option 1: Unzip the file to TOMCAT/webapps/source/ directory and change the source/WEB-INF/web.xml and other static html files like index.html to customize to your project.

• Option 2: Extract the web.xml file from source.war file

  unzip source.war WEB-INF/web.xml

  edit web.xml and re-package the jar file.

  zip -u source.war WEB-INF/web.xml

  Then copy the war files to TOMCAT/webapps directory.

• Option 3: Edit the Context container element for the webapp

  Copy source.war to TOMCAT/webapps

  When invoking OpenGrok to build the index, use -w <webapp> to set the context. If you change this(or set using OPENGROK_WEBAPP_CONTEXT) later, FULL clean reindex is needed.

  After the index is built, there's a couple different ways to set the Context for the servlet container:

  • Add the Context inside a Host element in TOMCAT/conf/server.xml

    <Context path="/<webapp>" docBase="source.war">
        <Parameter name="DATA_ROOT" value="/path/to/data/root" override="false" />
        <Parameter name="SRC_ROOT" value="/path/to/src/root" override="false" />
        <Parameter name="HEADER" value='...' override="false" />
    </Context>

  • Create a Context file for the webapp

    This file will be named <webapp>.xml.

    For Tomcat, the file will be located at: TOMCAT/conf/<engine_name>/<hostname>, where <engine_name> is the Engine that is processing requests and <hostname> is a Host associated with that Engine. By default, this path is TOMCAT/conf/Catalina/localhost or TOMCAT/conf/Standalone/localhost.

    This file will contain something like the Context described above.

To make ctags recognize additional symbols/definitions/etc. it is possible to specify configuration file with extra configuration options for ctags.

This can be done by setting OPENGROK_CTAGS_OPTIONS_FILE environment variable when running the OpenGrok shell script (or directly with the -o option for opengrok.jar). Default location for the configuration file in the OpenGrok shell script is etc/ctags.config under the OpenGrok base directory (by default the full path to the file will be /var/opengrok/etc/ctags.config).

Sample configuration file for Solaris code base is delivered in the doc/ directory.

OpenGrok script doesn't support this out of box, so you'd need to add it there. Usually to StdInvocation() function after line -jar ${OPENGROK_JAR}. It would look like this:

-A cs:org.opensolaris.opengrok.analysis.PlainAnalyzer

(this will map extension .cs to PlainAnalyzer) You should even be able to override OpenGroks analyzers using this option.

Both indexer and web app emit extensive log messages.

OpenGrok is shipped with the logging.properties file that contains logging configuration.

The OpenGrok shell script will automatically use this file if found under the base directory. It can also be set using the OPENGROK_LOGGER_CONFIG_PATH environment variable.

If not using the shell script, the path to the configuration file can be set using the -Djava.util.logging.config.file=/PATH/TO/MY/logging.properties java parameter.

You need to pass location of project file + the query to Search class, e.g. for fulltext search for project with above generated configuration.xml you'd do:

java -cp ./opengrok.jar org.opensolaris.opengrok.search.Search -R \
    /var/opengrok/etc/configuration.xml -f fulltext_search_string

For quick help run:

java -cp ./opengrok.jar org.opensolaris.opengrok.search.Search

You might need to modify the web application if you don't store the configuration file in the default location (/var/opengrok/etc/configuration.xml).

To configure the webapp source.war, look into the parameters defined in WEB-INF/web.xml of source.war (use jar or zip/unzip or your preferred zip tool to get into it - e.g. extract the web.xml file from source.war (unzip source.war WEB-INF/web.xml) file, edit web.xml and re-package the jar file (zip -u source.war WEB-INF/web.xml) ) file and change those web.xml parameters appropriately. These sample parameters need modifying(there are more options, refer to manual or read param comments).

• CONFIGURATION – the absolute path to XML file containing project configuration (e.g. /var/opengrok/etc/configuration.xml)
• ConfigAddress – port for remote updates to configuration, optional, but advised(since there is no authentication) to be set to localhost:<some_port> (e.g. localhost:2424), if you choose some_port below 1024 you have to have root privileges

If you need to change name of the web application from source to something else you need to use special option -w <new_name> for indexer to create proper xrefs, besides changing the .war file name. Be sure that when this changed you reindex cleanly from scratch. Examples below show just deploying source.war, but you can use it to deploy your new_name.war too.

Deploy the modified .war file in glassfish/Sun Java App Server:

• Option 1: Use browser and log into glassfish web administration interface: Common Tasks / Applications / Web Applications , button Deploy and point it to your source.war webarchive

• Option 2: Copy the source.war file to GLASSFISH/domains/YOURDOMAIN/autodeploy directory, glassfish will try to deploy it "auto magically".

• Option 3: Use cli from GLASSFISH directory:

  ./bin/asadmin deploy /path/to/source.war

  Deploy the modified .war file in tomcat: just copy the source.war file to TOMCAT_INSTALL/webapps directory.

Just run ant from command line in the top-level directory or use build process driven by graphical developer environment such as Netbeans.

Note: in case you are behind http proxy, use ANT_OPTS to download jflex, lucene. E.g.

ANT_OPTS="-Dhttp.proxyHost=?.? -Dhttp.proxyPort=80" ant

Run ant package to create package (specific for the operating system this is being executed on) under the dist/ directory.

Note: For full coverage report, a proper junit test environment is required. That would mean:

• You have to use Ant 1.9 and above
• At least junit-4.12.jar and its dependencies have to be in ant's classpath (e.g. in ./lib). The test task will download them automatically.
• Your PATH must contain directory with exuberant ctags binary
  • Note: make sure that the directory which contains exuberant ctags binary is prepended before the directory with plain ctags program.
• Your PATH variable must contain directories which contain binaries of appropriate SCM software which means commands hg, sccs, cvs, git, bzr, svn (svnadmin too). They must be available for the full report.

The tests are then run as follows:

ant -lib ./lib test

To check if the test completed without error look for AssertionFailedError occurrences in the TESTS-TestSuites.xml file produced by the test run.

If you want to run Findbugs (http://findbugs.sourceforge.net/) on OpenGrok, you have to download Findbugs to your machine, and install it where you have checked out your OpenGrok source code, under the lib/findbugs directory, like this:

cd ~/.ant/lib
wget http://..../findbugs-x.y.z.tar.gz
gtar -xf findbugs-x.y.z.tar.gz
mv findbugs-x.y.z findbugs

You can now run ant with the findbugs target:

$ ant findbugs
...
findbugs:
 [findbugs] Executing findbugs from ant task
 [findbugs] Running FindBugs...
 [findbugs] Warnings generated: nnn
 [findbugs] Output saved to findbugs/findbugs.html

Now, open findbugs/findbugs.html in a web-browser, and start fixing bugs!

If you want to install findbugs some other place than ~/.ant/lib, you can untar the .tar.gz file to a directory, and use the findbugs.home property to tell ant where to find findbugs, like this (if you have installed fundbugs under the lib directory):

ant findbugs -Dfindbugs.home=lib/findbug

There is also a findbugs-xml ant target that can be used to generate XML files that can later be parsed, e.g. by Jenkins.

If you want to check test coverage on OpenGrok, download jacoco from http://www.eclemma.org/jacoco/. Place jacocoagent.jar and jacocoant.jar in the opengrok/lib, ~/.ant/lib or into classpath (-lib option of ant).

Now you can instrument your classes and test them run:

ant -Djacoco=true -Djacoco.home=/<path_to>/jacoco jacoco-code-coverage

Now you should get output data in jacoco.exec

Look at jacoco/index.html to see how complete your tests are.

To check that your code follows the standard coding conventions, you can use checkstyle from http://checkstyle.sourceforge.net/

First you must download checkstyle from http://checkstyle.sourceforge.net/. You need Version 6.8 (or newer). Extract the package you have downloaded, and create a symbolic link to it from ~/.ant/lib/checkstyle, e.g. like this:

cd ~/.ant/lib
unzip ~/Desktop/checkstyle-7.6.zip
ln -s checkstyle-7.6 checkstyle

You also have to create symbolic links to the jar files:

cd checkstyle
ln -s checkstyle-7.6-all.jar checkstyle-all.jar

To run checkstyle on the source code, just run ant checkstyle:

ant checkstyle

Output from the command will be stored in the checkstyle directory.

If you want to install checkstyle some other place than ~/.ant/lib, you can untar the .tar.gz file to a directory, and use the checkstyle.home property to tell ant where to find checkstyle, like this (if you have installed checkstyle under the lib directory):

ant checkstyle -Dcheckstyle.home=lib/checkstyle

To check the quality of the OpenGrok code you can also use PMD from https://pmd.github.io/.

How to install:

cd ~/.ant/lib
unzip ~/Desktop/pmd-bin-5.5.4.zip
ln -s pmd-5.5.4/ pmd

To run PMD on the source code, just run ant pmd:

ant -Dpmd.home=~/.ant/lib/pmd pmd

Output from the command will be stored in the pmd subdirectory:

$ ls pmd
pmd_report.html  pmd_report.xml

If you want to install PMD some other place than ~/.ant/lib, you can unzip the .zip file to a directory, and use the pmd.home property to tell ant where to find PMD, like this (if you have installed PMD under the ./ext_lib directory):

ant pmd -Dpmd.home=ext_lib/pmd

To run CPD, just use the same as above, but use targets:

ant -Dpmd.home=ext_lib/pmd cpd cpd-xml

Which will result in:

$ ls pmd
cpd_report.xml cpd_report.txt

To see dependencies in the source code, you can use JDepend from https://github.com/clarkware/jdepend.

How to install:

cd ~/.ant/lib
unzip ~/Desktop/jdepend-2.9.1.zip
ln -s jdepend-2.9.1/ jdepend
cd jdepend/lib
ln -s jdepend-2.9.1.jar jdepend.jar

How to analyze:

ant jdepend

Output is stored in the jdepend directory:

$ ls jdepend/
report.txt  report.xml

Use a sonar runner with included sonar-project.properties properties, e.g. using bash:

cd <checkout_dir> # it has to contain sonar-project.properties!
export SONAR_RUNNER_OPTS="-Xmx768m -XX:MaxPermSize=256m"
export SERVERIP=10.163.26.78
~//Projects/sonar-runner-2.3/bin/sonar-runner \
    -Dsonar.host.url=http://${SERVERIP}:9000 \
    -Dsonar.jdbc.url=jdbc:h2:tcp://${SERVERIP}:9092/sonar

Travis depends on updated and working maven build. Please see .travis.yml, if your branch has this file, you should be able to connect your Github to Travis CI. OpenGroks Travis is here: https://travis-ci.org/OpenGrok/OpenGrok

The build can now be done through Maven (https://maven.apache.org/) which takes care of the dependency management and setup (calls Ant for certain actions).

You can test the code at the moment by running ./mvnw test which will execute all tests. Conditionally, if you don't have every type of repository installed, you can set it to unit-test only those which are found to be working on your system.

./mvnw test -Djunit-force-all=false

You can also force a specific repository test from running through the following system property

./mvnw test -Djunit-force-all=false -Djunit-force-git=true

While indexing big source repos you might consider using ZFS filesystem to give you advantage of datasets which can be flipped over or cloned when needed. If the machine is strong enough it will also give you an option to incrementally index in parallel to having the current sources and index in sync. (So Tomcat sees certain zfs datasets, then you just stop it, flip datasets to the ones that were updated by SCM/index and start tomcat again - outage is minimal, sources+indexes are always in sync, users see the truth)

OpenGrok script by default uses 2GB of heap and 16MB per thread for flush size of lucene docs indexing(when to flush to disk). It also uses default 32bit JRE. This might not be enough. You might need to consider this: Lucene 4.x sets indexer defaults:

DEFAULT_RAM_PER_THREAD_HARD_LIMIT_MB = 1945;
DEFAULT_MAX_THREAD_STATES = 8;
DEFAULT_RAM_BUFFER_SIZE_MB = 16.0;

• which might grow as big as 16GB (though DEFAULT_RAM_BUFFER_SIZE_MB shouldn't really allow it, but keep it around 1-2GB)

• the lucenes RAM_BUFFER_SIZE_MB can be tuned now using the parameter -m, so running a 8GB 64 bit server JDK indexer with tuned docs flushing(on Solaris 11):

  # export JAVA=/usr/java/bin/`isainfo -k`/java
  (or use /usr/java/bin/amd64/java )
  # export JAVA_OPTS="-Xmx8192m -server"
  # OPENGROK_FLUSH_RAM_BUFFER_SIZE="-m 256" ./OpenGrok index /source
  

Tomcat by default also supports only small deployments. For bigger ones you might need to increase its heap which might necessitate the switch to 64-bit Java. It will most probably be the same for other containers as well. For tomcat you can easily get this done by creating conf/setenv.sh:

# cat conf/setenv.sh
# 64-bit Java
JAVA_OPTS="$JAVA_OPTS -d64 -server"

# OpenGrok memory boost to cover all-project searches
# (7 MB * 247 projects + 300 MB for cache should be enough)
# 64-bit Java allows for more so let's use 8GB to be on the safe side.
# We might need to allow more for concurrent all-project searches.
JAVA_OPTS="$JAVA_OPTS -Xmx8g"

export JAVA_OPTS

For tomcat you might also hit a limit for http header size (we use it to send the project list when requesting search results):

• increase(add) in conf/server.xml

  maxHttpHeaderSize
  connectionTimeout="20000"
  maxHttpHeaderSize="65536"
  redirectPort="8443" />

  Refer to docs of other containers for more info on how to achieve the same.

The same tuning to Apache can be done with the LimitRequestLine directive:

LimitRequestLine 65536
LimitRequestFieldSize 65536

The initial index creation process is resource intensive and often the error java.io.IOException: error=24, Too many open files appears in the logs. To avoid this increase the ulimit value to a higher number.

It is noted that the hard and soft limit for open files of 10240 works for mid sized repositories and so the recommendation is to start with 10240.

If you get a similar error, but for threads: java.lang.OutOfMemoryError: unable to create new native thread it might be due to strict security limits and you need to increase the limits.

If multi-project search is performed frequently, it might be good to warm up file system cache after each reindex. This can be done e.g. with https://github.com/hoytech/vmtouch

The project has been originally conceived in Sun Microsystems by Chandan B.N.

• Chandan B.N, (originally Sun Microsystems)
• Trond Norbye, norbye.org
• Knut Pape, eBriefkasten.de
• Martin Englund, (originally Sun Microsystems)
• Knut Anders Hatlen, Oracle.
• Lubos Kosco, Oracle. http://blogs.oracle.com/taz/
• Vladimir Kotal, Oracle. http://blogs.oracle.com/vlad/
• Krystof Tulinger

Feel free to participate in discussion on the mailing lists:

[email protected] (user topics)

[email protected] (developers' discussion)

To subscribe, send email to <mailing_list_name>[email protected]