What is the best way to find out which revisions are in the current build at the start of the build
Hi I'm trying to create a readme file based on the commit messages for all the new revisions in the current build. Currently I do this by using the Jenkins API to request the changesets for the last build (e.g. we're currently building build #6 so the last build is #5). These changesets give me the last revision number for my VCS and then I can find out what has changed since. However this doesn't work quite that nicely because: * The first build doesn't have a change set, which seems a bit silly because obviously all the changes from all the commits until then are included * Sometimes a build fails at a point where there are no changes (e.g. failure to clear the workspace). Then this build has no change set * A build is run multiple times on a given VCS revision. So the question is what is the best / easiest way to determine which revisions are new in the current build in relation to the last successful build. Thanks Patrick
Re: What is the best way to find out which revisions are in the current build at the start of the build
On Thu, Apr 12, 2012 at 11:35 AM, Patrick petrikvanderve...@gmail.comwrote: Hi I'm trying to create a readme file based on the commit messages for all the new revisions in the current build. Currently I do this by using the Jenkins API to request the changesets for the last build (e.g. we're currently building build #6 so the last build is #5). These changesets give me the last revision number for my VCS and then I can find out what has changed since. However this doesn't work quite that nicely because: * The first build doesn't have a change set, which seems a bit silly because obviously all the changes from all the commits until then are included * Sometimes a build fails at a point where there are no changes (e.g. failure to clear the workspace). Then this build has no change set * A build is run multiple times on a given VCS revision. So the question is what is the best / easiest way to determine which revisions are new in the current build in relation to the last successful build. Thanks Patrick what about using the Jenkins api? parsing xml output of http://yourjenkins/job/yourjob/yourrevision/api/xml?xpath=//changeSet -- Ferenc Kovács @Tyr43l - http://tyrael.hu
Re: What is the best way to find out which revisions are in the current build at the start of the build
On Thu, Apr 12, 2012 at 12:44 PM, Ferenc Kovacs tyr...@gmail.com wrote: On Thu, Apr 12, 2012 at 11:35 AM, Patrick petrikvanderve...@gmail.comwrote: Hi I'm trying to create a readme file based on the commit messages for all the new revisions in the current build. Currently I do this by using the Jenkins API to request the changesets for the last build (e.g. we're currently building build #6 so the last build is #5). These changesets give me the last revision number for my VCS and then I can find out what has changed since. However this doesn't work quite that nicely because: * The first build doesn't have a change set, which seems a bit silly because obviously all the changes from all the commits until then are included * Sometimes a build fails at a point where there are no changes (e.g. failure to clear the workspace). Then this build has no change set * A build is run multiple times on a given VCS revision. So the question is what is the best / easiest way to determine which revisions are new in the current build in relation to the last successful build. Thanks Patrick what about using the Jenkins api? parsing xml output of http://yourjenkins/job/yourjob/yourrevision/api/xml?xpath=//changeSet but most importantly, why don't you fetch the changes of the current build? I mean the changes in the current build that interests you. The build will be created as soon as it starts, and when you buildstep kicks in, you can already have the changeset for your current build (and you have an environment variable which holds the current job number, so you have everything to fetch the changes for the current build imo. -- Ferenc Kovács @Tyr43l - http://tyrael.hu
Re: What is the best way to find out which revisions are in the current build at the start of the build
Hi That's what I'm doing at the moment. I'm currently calling http://yourjenkins/job/yourjob/yourrevision/api/xml?tree=changeSet[items[revision,msg]] which gives me an xml block with the revision and the message. However that only works if that build actually thinks it has changes (which it may not as per the reasons I provided). I guess the next step then is to find the previous build and try again? Also I was under the impression that the change set for the current build is only available after it is finished? On Thu, Apr 12, 2012 at 22:47, Ferenc Kovacs tyr...@gmail.com wrote: On Thu, Apr 12, 2012 at 12:44 PM, Ferenc Kovacs tyr...@gmail.com wrote: On Thu, Apr 12, 2012 at 11:35 AM, Patrick petrikvanderve...@gmail.com wrote: Hi I'm trying to create a readme file based on the commit messages for all the new revisions in the current build. Currently I do this by using the Jenkins API to request the changesets for the last build (e.g. we're currently building build #6 so the last build is #5). These changesets give me the last revision number for my VCS and then I can find out what has changed since. However this doesn't work quite that nicely because: * The first build doesn't have a change set, which seems a bit silly because obviously all the changes from all the commits until then are included * Sometimes a build fails at a point where there are no changes (e.g. failure to clear the workspace). Then this build has no change set * A build is run multiple times on a given VCS revision. So the question is what is the best / easiest way to determine which revisions are new in the current build in relation to the last successful build. Thanks Patrick what about using the Jenkins api? parsing xml output of http://yourjenkins/job/yourjob/yourrevision/api/xml?xpath=//changeSet but most importantly, why don't you fetch the changes of the current build? I mean the changes in the current build that interests you. The build will be created as soon as it starts, and when you buildstep kicks in, you can already have the changeset for your current build (and you have an environment variable which holds the current job number, so you have everything to fetch the changes for the current build imo. -- Ferenc Kovács @Tyr43l - http://tyrael.hu
Simple Dynamic ClearCase 1.2.2 generating nonsense commands
Hello, i use jenkins 1.459 with Simple Dynamic ClearCase plugin and i want to see the versions used for a build and also use jenkins to present differences in versions between builds. I configured the Simple Dynamic ClearCase plugin V1.2.2 in an example project to record the version information for a certain vob (lets say /vobs/example) and for the main branch. The command produced at start of the build by this plugin then is nonsense and fails to execute: cleartool setview -exec 'cleartool lshistory -branch main -last 10 - fmt %Nd| |%u| |%En| |%Vn| |%e| |%o| |%Nc\n -nco /vobs/example' example_view In my clearcase installation (7.1.1.1) cleartool does not know the command setview. Reading this line i think it is nonsense anyway, and that the program in fact just wanted to execute the inner part starting with cleartool lshistory Is there anybody out there that successfully used this plugin? Is it only the current version that is broken? How can i use jekins succesfully with clearcase to create version reports for each build that enable me to generate version difference reports between builds? Best regards Helmut
Maven builds with generic build project hang when test outputs text to stderr
Hello, as posted in a previous message i switched from using maven builds to generic builds with maven build steps. These projects always hang at the end when the tests executed produce output to stderr. As i dont know a way to suppress this output (and in fact i do want to see it after the build), how can i enable jenkins not to stumble over this output? One other strange thing with this output is, that it appears not in the context of the test case, but at the very end of the build (after the line Tests run:), as if the output to stderr during the test was held back by someone. Best regards Helmut
Re: Warnings Plugin and Mail-ext Integration
I'm having a few difficulties on this. The GroovyShell is initialized
with a ClassLoader like this:
private static final class Script implements
DelegatingCallableString,RuntimeException {
private final String script;
private transient ClassLoader cl;
private Script(String script) {
this.script = script;
cl = getClassLoader();
}
public ClassLoader getClassLoader() {
return Jenkins.getInstance().getPluginManager().uberClassLoader;
}
public String call() throws RuntimeException {
// if we run locally, cl!=null. Otherwise the delegating
classloader will be available as context classloader.
if (cl==null) cl =
Thread.currentThread().getContextClassLoader();
GroovyShell shell = new GroovyShell(cl);
StringWriter out = new StringWriter();
PrintWriter pw = new PrintWriter(out);
shell.setVariable(out, pw);
try {
Object output = shell.evaluate(script);
if(output!=null)
pw.println(Result: +output);
} catch (Throwable t) {
t.printStackTrace(pw);
}
return out.toString();
}
}
The part I am interested in is when it uses the uberClassLoader vs
using the getContextClassLoader. How is cl set to null for executing
remotely? I am trying to replicate this setup for the groovy
templating in the email-ext plugin so it can load other plugins and
such.
Thanks,
slide
On Wed, Apr 11, 2012 at 3:05 PM, Darren Syzling dsyzl...@gmail.com wrote:
Thanks for looking into this - replying late but yes I tried explicit
imports to no avail. Makes sense that a different class loader was
being used.
Look forward to testing an updated version when available.
Darren
On 11 April 2012 21:06, Slide slide.o@gmail.com wrote:
Ok, it looks like the groovy script console sets the classloader for
the context to Jenkins.getInstance().getPluginManager().uberClassLoader,
which the email-ext plugin does NOT do. I will look at updating this
and trying it out and if it works, I'll release a new version of
email-ext.
slide
On Wed, Apr 11, 2012 at 8:51 AM, Slide slide.o@gmail.com wrote:
I also can't seem to get the groovy template to load the
WarningsResultAction class. I tried similar things to what you did
below. I will see if I can figure out what this is the case.
Thanks,
slide
On Wed, Apr 11, 2012 at 5:57 AM, Darren Syzling dsyzl...@gmail.com wrote:
Ulli,
I'd be happy to work on something together. I did however try a few
variants but the hudson.plugins.warnings.WarningsResultAction class
could not be found, I tried variants of this:
warningsResultAction =
build.getAction(Class.forName(hudson.plugins.warnings.WarningsResultAction))
warningsResultAction =
build.getAction(hudson.plugins.warnings.WarningsResultAction.class)
and using the class loader:
ClassLoader cl = it.class.getClassLoader()
def warningsClass =
cl.loadClass(hudson.plugins.warnings.WarningsResultAction, true)
Is there something different about the groovy parser engine loaded by
email-ext which would mean other plugin classes were not available.
Unfortunately I then decided to upgrade to Jenkins 1.459 along with:
Static analysis utilities 1.38
Warnings 3.28
Static analysis collector 1.24
Email-ext 2.18
This has broken my ability to build and configure the project, when I
build:
11-Apr-2012 13:55:43 hudson.model.Executor run
SEVERE: Executor threw an exception
java.lang.AssertionError: class
hudson.plugins.warnings.WarningsPublisher is missing its descriptor
at jenkins.model.Jenkins.getDescriptorOrDie(Jenkins.java:1076)
at hudson.tasks.Publisher.getDescriptor(Publisher.java:123)
at hudson.tasks.Recorder.getDescriptor(Recorder.java:51)
at
hudson.plugins.warnings.WarningsPublisher.getDescriptor(WarningsPublisher.java:323)
at
hudson.plugins.warnings.WarningsPublisher.getDescriptor(WarningsPublisher.java:41)
at hudson.model.Descriptor.toMap(Descriptor.java:873)
at hudson.util.DescribableList.toMap(DescribableList.java:128)
at hudson.model.Project.getPublishers(Project.java:109)
at hudson.model.Build$RunnerImpl.cleanUp(Build.java:171)
at hudson.model.Run.run(Run.java:1457)
at hudson.model.FreeStyleBuild.run(FreeStyleBuild.java:46)
at
hudson.model.ResourceController.execute(ResourceController.java:88)
at hudson.model.Executor.run(Executor.java:238)
and when I try and configure the project I can't access email-ext or
warnings configuration information:
11-Apr-2012 13:46:50 hudson.ExpressionFactory2$JexlExpression evaluate
WARNING: Caught exception evaluating: i.descriptor. Reason:
java.lang.reflect.InvocationTargetExcept
ion
java.lang.reflect.InvocationTargetException
at
Authentification failed when deploying a project
Hello, I am new using Jenkins, and am having some problems when I try to deploy a project in a remote machine: [INFO] [ERROR] BUILD ERROR [INFO] [INFO] Error retrieving previous build number for artifact 'host:project:jar': repository metadata for: 'snapshot host:project: 1.0-SNAPSHOT' could not be retrieved from repository: snapshot due to an error: Authentication failed: The host was not known and was not accepted by the configuration: host How could I set up this information?? Thanks in advance
Re: how to get access to the data for the job being promoted?
On 12-04-10 04:54 PM, Sami Tikka wrote: You do not exactly come clean on what kind of data it is you are looking for. Well, specifically, I was looking to know in the executed job which job was being promoted. If you want access to promoted build artifacts, Copy Artifacts plugin gives that in the triggered build in Specific build - Last promoted This sounds racy, but moot in any case since there is no mention of anything to do with promoted builds in the list of artifacts available for copying. But that's ok, see below... If it is something else you want, you could maybe send it over as a parameter. Yes, this is what I was looking for! Use the parameterized trigger in the promotion action to send what ever data you want to pass to the triggered job. Yeah. I hadn't realized that any action plugin that I add would be available as a promotion action. Simply installing the parameterized trigger plugin I was able to send the PROMOTED_* variables to the downstream job. Thanks much! b. signature.asc Description: OpenPGP digital signature
serializing actions in a promotion process
I have a promotion configuration using the Promoted Builds Plugin. This configuration includes the building of two other jobs in it's Actions list, specifically two Trigger parameterized build on other projects items. These seem to happen in parallel though and I would like to force them to happen serially. Additionally, I would like to copy the build artifacts of the first one run in the second one. I think I would be able to do that if the new build number of the first executed project was added to the environment of process running the actions, so then I could pass it to the second job as a parameter. One way of achieving this is of course to combine the two jobs into a single one but the reality is that one of the jobs is called elsewhere so combining it's steps into this first job would be a duplication and future synchronization nightmare. Any ideas? Cheers, b. signature.asc Description: OpenPGP digital signature
Re: Promoted Builds Plugin: Does not automatically promote despite qualifications being met
It looks like a weird bug. It happens on one set of jobs but not on another one I have. Another problem I'm having is that when you manually promote a build that triggers a downstream build, the triggered downstream build does not include information from the upstream project, which is a big problem. On Thu, Apr 12, 2012 at 4:05 AM, Taubert, Helge t...@zuehlke.com wrote: Hi! ** ** We have a similar issue for some versions back now. We get the following warning: ** ** ** ** WARNING: Job_A appears to use this job as a promotion criteria, but no fingerprint is recorded. Fingerprint needs to be enabled on both this job and Job_A ** ** If you check the fingerprints manually, everything is fine, but the promotion does not work. It worked one and a half year ago. Then there were some race condition which made it fail sometimes. These were fixed and it worked for a while. Then it stopped working at all. ** ** It would be nice to have this feature working again. ** ** Greetings, ** ** Helge ** ** *From:* jenkinsci-users@googlegroups.com [mailto: jenkinsci-users@googlegroups.com] *On Behalf Of *Nicky Ramone *Sent:* Mittwoch, 11. April 2012 18:31 *To:* jenkinsci-users@googlegroups.com *Subject:* Promoted Builds Plugin: Does not automatically promote despite qualifications being met ** ** Hello. ** ** I have the following build sequence with Maven jobs: A -- B -- C and I want to promote A when C has finished successfully. I am tracking fingerprints of the same artifact in all 3 jobs, and I verify that they are correct. ** ** When C finishes, the Promotion Status looks fine with met qualifications: ** ** Promotions my_promotion [Force promotion] This promotion has not happened. Met Qualification *Downstream builds succeeded* C #8 http://localhost:8080/job/C/8/ Unmet Qualification ** ** Despite that, the promotion never occurs (you can see the label This promotion has not happened). I found a couple of threads/issues that seem to be pointing to the same problem, but no solution available: http://jenkins.361315.n4.nabble.com/Promotion-is-not-working-td3643188.html https://issues.jenkins-ci.org/browse/JENKINS-10199 ** ** I'm using Jenkins 1.455 and Promoted Builds Plugin 2.4 Has anyone else come across this issue? ** ** Thanks Cheers. ** **
Re: pre-build/tested commits
On 2012-04-11 21:47, Sami Tikka wrote: Probably not. AFAIK there is no way to tell git this branch is now ready and merged and it will never again receive new commits and it can be deleted It is always possible to checkout an old branch and add new commits to it. If you are concerned about the large number of branches in the central repository, you could tell each developer they are allowed to have only 1 branch in there. A developer would always need to merge the feature branch he is working on to his own branch on the central repo. OTOH I'm not sure why the large number of branches would be a problem? Can you explain it? I really can't. We are just now starting up with git. Going from CVS to git is sort of like going from Visual Basic to C++ and I worry that we are going to get our feet shot off. Well, I worried. This quote calmed me down :-) It is easy to shoot your foot off with git, but also easy to revert to a previous foot and merge it with your current leg. —Jack William Bell Thanks for your help, Thomas
