Jenkins/Hudson job push changes to Mercurial - ant

I have a Jenkins/Hudson that increase some version numbers of some modules in preparation to publish a NetBeans update center to Sourceforge. I'm trying to push those changes into the repository using ant4hg ant task and I'm having some issues. here's that part of the task relevant to this:
<change files here>
<hg cmd="commit" dir="." message="${message}" user="${hg.user}">
<dirset dir="./">
<include name="**/*.mf"/>
</dirset>
</hg>
<hg cmd="push" dir=".">
<auth user="${hg.user}" password="${hg.password}" />
</hg>
You can see the related bug here. Anyway, is there a better way to do this from Jenkins/Hudson? Thanks in advance!
Edit
I'm trying to push a change and it doesn't work. I tried to use the command as specified in the documentation:
<hg cmd="push" dir=".">
<auth user="${hg.user}" password="${hg.password}" />
</hg>
But I get:
$ hg push
pushing to javydreamercsw#bitbucket.org/javydreamercsw/simple-marauroa-rcp">https://javydreamercsw#bitbucket.org/javydreamercsw/simple-marauroa-rcp [javydreamercsw#bitbucket.org/javydreamercsw/simple-marauroa-rcp" target="_blank">^]
pushing subrepo Card Game Interface to javydreamercsw#bitbucket.org/javydreamercsw/card-game-interface">https://javydreamercsw#bitbucket.org/javydreamercsw/card-game-interface [javydreamercsw#bitbucket.org/javydreamercsw/card-game-interface" target="_blank">^]
searching for changes
no changes found
pushing subrepo Simple Marauroa Java to https://bitbucket.org/javydreamercsw/simple-marauroa-java [^]
searching for changes
no changes found
pushing subrepo jWrestling to https://bitbucket.org/javydreamercsw/jwrestling [^]
searching for changes
no changes found
searching for changes
[ERROR] hg > abort: http authorization required
Result: 255
[ERROR] FAIL TO EXECUTE : hg push (255)
Z:\NetBeans\Simple Marauroa\Marauroa Application Manager\build.xml:680: FAIL TO EXECUTE : hg push (255)
If I do it from NetBeans and/or TortoisHg it works. I thought I provided the wrong credentials but I know they are correct.
Previous to that command I do a commit and I get this output:
$ hg commit --include */.mf --user xxxxx --message "Update versions for dailybuilds" Z:\NetBeans\Simple Marauroa\Marauroa Application Manager
I pass the same user to the push command adding the correct password as well.

Related

Ivy retrieve latest artifact with pre release

I have an ant task that uses ivy:retrive task with revision set to something like "1.2.3+". My problem is that I have artifacts which have the same first three digits but end differently.
Artifacts sample:
1.2.2-b1-hash
1.2.3-b2-someHash
1.2.3-b3-anotherHash
1.2.3-b4-someOtherHash
1.2.4-b5-latestVersion
Currently I have the following target:
<target name="retrieve-apps">
<ivy:retrieve
inline="true"
conf="core"
organisation="com.myOrg.myApp"
module="myApp"
revision="1.2.3+"
pattern="${basedir}/[artifact].[ext]" />
</target>
When I run the target, I always get the artifact "1.2.3-b2-someHash". Is there a way that I can change the revision or pattern to get the artifact "1.2.3-b4-someOtherHash" while preserving a rule for the first three digit ?
I have tried to use the latest.integration as revision but it returns "1.2.4-b5-lastestVersion" which is not what I want. I have also tried to use "1.2.3-b3-someOtherHash" but I will have to change it every time I want to upgrade to the latest build of that version.

Permission denied while running maven command jenkins

I am facing permission denied issues while running a maven clean test on a pom.xml present in /home/tes/pom.xml from jenkins. Any suggestions
Building in workspace /var/lib/jenkins/workspace/jenkins
[jenkins] $ /usr/share/maven/bin/mvn -f /home/tes/pom.xml clean test
[INFO] Scanning for projects...
[ERROR] The build could not read 1 project -> [Help 1]
[ERROR]
[ERROR] The project (/home/tes/pom.xml) has 1 error
[ERROR] Non-readable POM /home/tes/pom.xml: /home/tes/pom.xml (Permission denied)
[ERROR]
[ERROR] To see the full stack trace of the errors, re-run Maven with the -e switch.
[ERROR] Re-run Maven using the -X switch to enable full debug logging.
[ERROR]
[ERROR] For more information about the errors and possible solutions, please read the following articles:
[ERROR] [Help 1] http://cwiki.apache.org/confluence/display/MAVEN/ProjectBuildingException
Build step 'Invoke top-level Maven targets' marked build as failure
Finished: FAILURE
Just grant read access for all to the pom.xml file:
chmod +r /home/tes/pom.xml
The moderator has deleted my previous answer; I know that it has passed some time since this first post, but I ended up exactly in the same place. I insist in complementing all the answers because I've found what was the problem in my case,having the same result as reported in this case.
I'd like to comment that, in my case, the reason was that I did not have JDK, but just JRE configured in JAVA_HOME of the Global Tool Configuration > JDK > JDK Installation. As a matter of fact Jenkins warns you that the path is not pointing to something containing a JDK.
JDK is needed for some Plugins, like Maven Integration or Docker Pipeline, as Java requirements section in Jenkins documentation states. So this must checked because it surprisingly results in the weird error:
> git rev-parse refs/remotes/origin/celsus_0-1-0^{commit} # timeout=10
Checking out Revision 0c533cf2327a416a254afa2348abafe7790ba67f (refs/remotes/origin/celsus_0-1-0)
> git config core.sparsecheckout # timeout=10
> git checkout -f 0c533cf2327a416a254afa2348abafe7790ba67f # timeout=10
Commit message: "first embrio of container image"
> git rev-list --no-walk 0c533cf2327a416a254afa2348abafe7790ba67f # timeout=10
Parsing POMs
Established TCP socket on 40251
[build] $ /usr/lib/jvm/bin/java -cp /var/lib/jenkins/plugins/maven-plugin/WEB-INF/lib/maven35-agent-1.13.jar:/usr/share/maven/boot/plexus-classworlds-2.x.jar:/usr/share/maven/conf/logging jenkins.maven3.agent.Maven35Main /usr/share/maven /var/cache/jenkins/war/WEB-INF/lib/remoting-4.5.jar /var/lib/jenkins/plugins/maven-plugin/WEB-INF/lib/maven35-interceptor-1.13.jar /var/lib/jenkins/plugins/maven-plugin/WEB-INF/lib/maven3-interceptor-commons-1.13.jar 40251
ERROR: Failed to parse POMs
java.io.IOException: Cannot run program "/usr/lib/jvm/bin/java" (in directory "/var/lib/jenkins/workspace/build"): error=2, No such file or directory
at java.base/java.lang.ProcessBuilder.start(ProcessBuilder.java:1128)
at java.base/java.lang.ProcessBuilder.start(ProcessBuilder.java:1071)
at hudson.Proc$LocalProc.<init>(Proc.java:252)
at hudson.Proc$LocalProc.<init>(Proc.java:221)
at hudson.Launcher$LocalLauncher.launch(Launcher.java:996)
at hudson.Launcher$ProcStarter.start(Launcher.java:508)
at hudson.maven.AbstractMavenProcessFactory.newProcess(AbstractMavenProcessFactory.java:280)
at hudson.maven.ProcessCache.get(ProcessCache.java:236)
at hudson.maven.MavenModuleSetBuild$MavenModuleSetBuildExecution.doRun(MavenModuleSetBuild.java:802)
at hudson.model.AbstractBuild$AbstractBuildExecution.run(AbstractBuild.java:514)
at hudson.model.Run.execute(Run.java:1907)
at hudson.maven.MavenModuleSetBuild.run(MavenModuleSetBuild.java:543)
at hudson.model.ResourceController.execute(ResourceController.java:97)
at hudson.model.Executor.run(Executor.java:429)
Caused by: java.io.IOException: error=2, No such file or directory
at java.base/java.lang.ProcessImpl.forkAndExec(Native Method)
at java.base/java.lang.ProcessImpl.<init>(ProcessImpl.java:340)
at java.base/java.lang.ProcessImpl.start(ProcessImpl.java:271)
at java.base/java.lang.ProcessBuilder.start(ProcessBuilder.java:1107)
... 13 more
Finished: FAILURE
Similar case as originated this case, as the different nature of the solutions proposed points out. I'd like to bring the attention that these errors having to do with permissions, or parsing of the pom might hide the real root cause, and error messages are not always as explicit and clear as they should be.
Hope it helps others bumping into the same issue.
I installed maven plugin for jenkins and provided the fully qualified path of pom.xml in root pom.xml and it fixed the issue
I think the reason this happens is maven has the right to run pom.xml ie. mvn command can be run through terminal but not through jenkins.
So we will need to add the current user to jenkins for it to work.
Follow this blogpost. It worked for me:
blogpost

Configure Jenkins to deploy PHP project that passed PHPUNit

I have PHP Project, that is hosted on GitHub.
Now, I'd like to configure Jenkins to run unit tests so that:
Whenever developer push/commits code to specific branch, it triggers corresponding PHPUnit build job.
If commit passes the unit tests, the source code is deployed (assuming I already have the required script to deploy).
The question is how to trigger the deployment script when source code passes the unit test (i.e. PHPUnit tests succeed)?
Please suggest to me the way to do that, which plugin I should try to achieve the result?
Thanks!
This is going to be a long post, as there's a lot involved, but it works a treat:
You will need:
Ant
Git Publisher plugin
Ant and phpunit will need to be on your PATH
Step 1: Configure your project
In your Jenkins, configure your project to 'Poll SCM' under the Git option. Leave the 'Schedule' as blank. Under 'branches to build' set that as the branch you want to build your release package from.
Reference:
Step 2: Run ant for every build
Add a build step to 'Invoke Ant'
If you don't use Ant already, create a build.xml file in your project root, add it to Git and have the following contents:
<?xml version="1.0" encoding="UTF-8"?>
<project default="full-build">
<property name="phpunit" value="phpunit"/>
<target name="full-build"
depends="phpunit-unittests,-check-failure"
description="runs the tests"/>
<target name="phpunit-unittests"
unless="phpunit-unittests.done"
description="Run unit tests with PHPUnit">
<exec executable="cmd" failonerror="true" resultproperty="result.phpunit" taskname="phpunit-unittests">
<arg value="/c"/>
<arg value="${phpunit}"/>
<arg value="--configuration"/>
<arg path="${basedir}/phpunit.xml"/>
<arg value="--testsuite=Unit"/>
</exec>
<property name="phpunit-unittests.done" value="true"/>
</target>
<target name="-check-failure">
<fail message="PHPUnit did not finish successfully">
<condition>
<not>
<equals arg1="${result.phpunit}" arg2="0"/>
</not>
</condition>
</fail>
</target>
</project>
That will run all unit tests whenever the Ant task is invoked, which is now set for every time the project is built.
Then, install the Git Publisher tool. Configure as follows:
This creates a new release tag upon a successful build. You will use this later to publish the release to the final location. Note: There are different variables that Git Publisher provides for use, commit hash, user etc so use what you want. I stick to an incremental tag of v1.1.BUILD as that's a bit more standard.
Lastly, you will need to add a Git hook which will trigger a build upon a commit/push from any location.
Navigate to your repository folder and within that the 'hooks' directory.
Create a new file named 'post-receive' (you will see examples in there; overwrite this one). Place the following content in:
#!/bin/bash
while read oldrev newrev refname
do
branch=$(git rev-parse --symbolic --abbrev-ref $refname)
if [ "master" == "$branch" ]; then
curl http://YOUR_JENKINS_URL:8080/git/notifyCommit?url=YOUR_GIT_REPOSITORY_URL
fi
done
That should do the job nicely. I have left out implementation details of how you actually release your project as everyone does this differently. There are options to FTP files to a location, and all sorts. Personally I go into the folder where the application resides and do a checkout of the newly created tag - a one line command. Whatever suits your environment.
Other stuff I've ommitted but you will find useful - the Ant build task can do literally anything - In mine, I run composer to install dependences, run bower, run grunt, do syntax checking, coding standard checking, fire up selenium and run web tests, and a load of other stuff. It's a perfect combination of tools to automate the whole project deployment.

CruiseControl.net ndk-build on Windows 64bit without Cygwin

Latest Android NDK (version 8) allows to call ndk-build without additional cygwin installed.
My question is: can I plug this into CruiseControl.Net for periodical native build?
I assume that most likely I would need to use Ant as a build method and then add it to ccnet server config.
So the questions are:
How to call ndk-build.cmd from an Ant command
How to configure build.xml file?
I tried several approaches, but all of those failed. Could you give me some clues if it is possible at all?
I ran into the same problem (the CreateProcess failures, etc) when trying to invoke "ndk-build.cmd" directly from build.xml, using CC.net on Windows. I figured out how to get this to work, and so I felt compelled to share my solution.
First, create a 1-line Windows command file as follows (I named it "ndkwrapper.cmd"):
sh ndkwrap.sh %1 %2 %3
Next, create a simple bash shell script as follows (named "ndkwrap.sh"):
#!/usr/bin/sh
# Wrapper around ndk-build script for Windows
#
NDK_PROJECT_PATH=/cygdrive/c/workspace/agnc_android
export NDK_PROJECT_PATH
/cygdrive/c/Android/android-ndk-r8b/ndk-build $1 $2 $3
Of course, your exact paths (for your workspace and NDK directories) may be different, but note, in particular, that "/cygdrive/c" refers to "C:".
One very important step here, after you create the script above, is to convert the line endings from Windows to UNIX. You can do this with a utility called "dos2unix.exe" which is freely available. I don't remember where I got it, but it was part of some open source package of GNU/Linux tools ported to Windows. (Google on "UNIX file utilities for Windows".) If you don't convert the line endings, then sh or bash will have trouble reading the script, and you'll get all kinds of erros about "\r" ...
So, to invoke the equivalent of "ndk-build.cmd clean", for example, you'd type "ndkwrapper.cmd clean" to delete your intermediate and output NDK-related build files, in preparation for a full NDK build.
In your build.xml file for CC.net on Windows, you can invoke the NDK makefile as follows:
<tasks>
<exec>
<executable>cmd.exe</executable>
<baseDirectory>C:\Android</baseDirectory>
<buildArgs>/C ndkwrapper.cmd clean</buildArgs>
</exec>
Hope this helps!
Ben
i observed problems with running ndk-build as an CCNET task as well.
It took me a while, but at the end i noticed, that you have to provide HOST_OS and HOST_ARCH as ndk-build parameters to let it run.
<exec>
<executable>cmd</executable>
<buildArgs>/C ndk-build HOST_OS=windows HOST_ARCH=x86 -B NDK_PROJECT_PATH=[PROJECT] APP_BUILD_SCRIPT=[ANDROIDMKFILE] NDK_APPLICATION_MK=[APPLICATIONMKFILE] NDK_LOG=1</buildArgs>
<buildTimeoutSeconds>120</buildTimeoutSeconds>
</exec>
hope it helps anyone to save time.
Ok I got some progress, I am able to build the jni code via ant or nant but in both cases plugging it to the cc.net server config gives me strane error:
but now CC.net gives me strange errors:
<builderror>
<type>NAnt.Core.BuildException</type>
<message><![CDATA[External Program Failed: G:\\android-ndk-r8b\\ndk-build.cmd (return code was 2)]]></message>
<location>
<filename>G:/MYPath/project.build</filename>
<linenumber>7</linenumber>
<columnnumber>4</columnnumber>
</location>
<stacktrace><![CDATA[ at NAnt.Core.Tasks.ExternalProgramBase.ExecuteTask()
at NAnt.Core.Tasks.ExecTask.ExecuteTask()
at NAnt.Core.Task.Execute()
at NAnt.Core.Target.Execute()
at NAnt.Core.Project.Execute(String targetName, Boolean forceDependencies)
at NAnt.Core.Project.Execute()
at NAnt.Core.Project.Run()]]></stacktrace>
</builderror>
</failure>
<duration>296.40000000000003</duration>
</buildresults>Buildfile: file:///G:/MYPath/project.build
Target framework: Microsoft .NET Framework 4.0
Target(s) specified: build
clean:
[echo] Starting Android ndk CLEAN ...
[exec] Android NDK: Unable to determine HOST_OS from uname -s:
[exec] Android NDK: Please define HOST_OS in your environment.
[exec] process_begin: CreateProcess(NULL, uname -s, ...) failed.
[exec] G:/android-ndk-r8b/build/core/init.mk:131: *** Android NDK: Aborting. . Stop.
BUILD FAILED - 0 non-fatal error(s), 2 warning(s)
My project in cc.net config:
<project>
<name>MY_PROJECT_NAME</name>
<workingDirectory>PATH_TO_MY_PROJECT</workingDirectory>
<tasks>
<nant>
<executable>G:\nant-0.92\bin\Nant.exe</executable>
<baseDirectory>PATH_TO_MY_PROJECT</baseDirectory>
<buildFile>MYPROJECTNAME.build</buildFile>
<buildArgs>build</buildArgs>
<buildTimeoutSeconds>1200</buildTimeoutSeconds>
</nant>
</tasks>
</project>
And my NAnt build file:
<project name="my name" default="build" basedir=".">
<description>My project Build Files.</description>
<target name="clean" description="remove all generated files">
<echo message="Starting Android ndk CLEAN ..."/>
<exec program="G:\\android-ndk-r8b\\ndk-build.cmd" failonerror="true"
basedir="MY PROJECT PATH">
<arg value="clean" />
</exec>
<echo message="Android ndk CLEAN done."/>
</target>
<target name="build" description="remove all generated files" depends="clean">
<echo message="Starting Android ndk BUILD ..."/>
<exec program="G:/android-ndk-r8b/ndk-build.cmd" failonerror="true" />
<echo message="Android ndk BUILD done."/>
</target>
</project>
As I said I can run a Nant.exe for my project and it cleans and build correctly.
It looks like the cc.net tries to run the other ndk-build commend which is used for linux and is missing some commands as uname.
Do you have any idea what cI could be doing wrong ?
The original question is asking about ant builds. This answer is related to a problem in CC.NET and ant and gradle builds are going to be affected in the same way.
We are using gradle and with some custom gradle tasks it is possible to compile the native code of your project as well by calling ndk-build.cmd (from a path that has no spaces in it).
After the gradle file is prepared properly shell initiated gradle builds will work but cc.net initiated builds will fail.
Our CC.NET task is defined as follows:
<exec executable=".\gradlew.bat">
<baseDirectory>$(projSrcDir)</baseDirectory>
<buildArgs>clean assemblePlayRelease assembleProRelease</buildArgs>
<buildTimeoutSeconds>900</buildTimeoutSeconds>
</exec>
The problem is related to a CC.NET bug that causes the shell variable names to be in lower case. Windows shell scripts do not care about the case of shell variable names but because the NDK fires up make variable case becomes an issue. The init.mk make file that is part of the build process relies on shell variable names to identify the host OS type. In particular the shell variable OS. Normally the OS value is defined like this:
OS=Windows_NT
But when the variable is passed to gradle from CC.NET it gets transformed into this:
os=Windows_NT
This causes the detection mechanism in init.mk to fail and tries to execute the unix detection sequence and looks for the Unix uname utility that is not present on Windows resulting in:
Android NDK: Unable to determine HOST_OS from uname -s:
Android NDK: Please define HOST_OS in your environment.
make.exe: Entering directory `D:/CC.NET/svncheckout/pes/src/project/src/main/jni'
make.exe: Leaving directory `D:/CC.NET/svncheckout/pes/src/project/src/main/jni'
And ultimately:
process_begin: CreateProcess(NULL, uname -s, ...) failed.
c:/ndk/build/core/init.mk:160: *** Android NDK: Aborting. . Stop.
FAILURE: Build failed with an exception.
The workaround is less then ideal but it gets the job done. The idea is to edit the ndk-build.cmd and change the case of the passed variables only when we are building from CC.NET.
Simply insert this code after the #echo off line in ndk-build.cmd:
IF DEFINED ccnetproject (
ECHO Applying Environment variable hack for CC.NET.
SET OS2=%os%
SET OS=
REM SET PATH=%path%
SET TEMP2=%temp%
SET TEMP=
)
IF DEFINED ccnetproject (
SET OS=%OS2%
SET OS2=
SET TEMP=%TEMP2%
SET TEMP2=
)
This script first makes temporary copies of values in the lower case OS and TEMP variables. Next it undefines them by clearing their values. An finally the reverse is done.
Two steps are needed because just saying
SET OS=%os%
doesn't do much since scripts are case insensitive. It first locates the OS variable, finds a lower case version and assigns its own value back to itself.
This was tested on CC.NET v1.6.7981.1, gradle v1.1.0 and Android NDK v10 (64bit).

Ant fails with liquibase path

I am trying to get phing to work nice with liquibase. But pPing gives this illustrous error (which I honestly can't find online)
Execution of target "update-database" failed for the following reason: PathElement (unknown) doesn't support the 'location' attribute.
BUILD FAILED
PathElement (unknown) doesn't support the 'location' attribute.
Total time: 0.1206 seconds
the specific commands for that are:
<path id="liquibasepath">
<pathelement location="${basedir}/install/lib/liquibase.jar" />
<pathelement location="${basedir}/install/lib/jdbc-mysql.jar" />
</path>
I tried searching online but there is no good documentation for using phing with liquibase. and the error's neither turn up decent results.
Try using the "path" attribute in pathelement instead of location
The correct answer should be: This is ant, not phing. I was trying to run it with the wrong program. (there really should be some indication of this in build files ;)).
The person who commented got me on hte right track, can't accept that as an answer tho.

Resources