My question may be silly but I've been trying several ways and I still can't do what I want, i.e.:
use the scp target of Ant to target a remote machine and execute
a script there
this script creates a dynamic list of files
get this list of files (only their names) back in Hudson to use it in the next build step (another scp from Ant)
I tried to use environment variables but they are interpreted by Hudson so I'm stuck here...
Globally my question would be: how to get a result from an Ant build step ?
Thanks for your ideas,
Emmanuel
You may find File parameter useful. This allows you to create an input file, pass it to build. You may need to write script/ant script to process the file though.
In the long term you may evaluate a Hudson farm. This will allow to create tasks that span multiple machines , pass results around. (https://wiki.jenkins-ci.org/display/JENKINS/Plugins)
You can get the ID(s) of the job that triggered your job via the API and fetch their status.
Related
Is there a way to overwrite a value contained within a config.properties file via Jenkins?
I have the following config.properties file contained within my automation framework:
browser=chrome
url=http//www.example.com
If the value of chrome get changed to firefox then all tests will now execute within firefox browser.
I can manually change this value by directly accessing the config.properties file but can the value get altered via jenkins?
I use the Pipeline Utility Steps plugin to read properties files, and it looks like it can write a few other types of files, but not properties files.
It seems to me that you want to make this change in this file so you can run some tests first in one browser, then in another. If this is the case, I think a better way to handle this is to try to get your tests to point to different files. This is a little cleaner, and allows things like parallel execution and when you find that another thing needs to change in the future, you won't be writing so many things to the file in a script, which gets a little error prone.
If you can't make your tests execute against a different properties file, you could have a copy of each file you need, and then copy them to them appropriate filename to execute your tests.
But maybe I made poor assumptions as to your setup here. ;)
Yes.
You can create a build parameter as $browser to accept the value say "firefox" and using sed inside "execute shell", replace the value in config.properties.
Once done, execute your scripts.
This is just overview as you have not posted details about your config.properties file, its location, if you are using Jenkins jobs or jenkinsfile/pipeline etc.
I'd like to run several builds concurrently in Jenkins for the same job. I run at maximum 3 builds concurrently. I want each build to run with a parameter that must be unique from a pool for parameters. For instance, pool=[1, 2, 3]: The 1st build picks "1", 2nd picks "2" and the 3rd picks "3".
I must ensure that different builds can't pick the same parameter.
After building, the parameter is available again.
How can I do it?
Alternative: How can I count the number of builds running in this project and pass it as parameter?
At first, select the checkbox button named build-concurrently-if-neccesary to ensure the same job could build concurrently. you'd better read the help-html seriously before
The isolated environments for building different jobs make that data could not be shared each other in a simple way.
Here is a solution that trigger the buildWithParameters link by jenkins rest api to control the pool in the program procedure of your own.
add a string-parameter in job's config.
post the string parameter to http://$JENKINS_SERVER_URL/job/$JOB_NAME/buildWithParameters
Maybe it's the most convenient way if no available plugin found.
I found a plugin in github and asked the author to publish it. It works well and solves my problem.
Jenkins Parameter Pool Plugin
I am trying to find a way to prompt the user to select and upload a generic file from a local machine to a Jenkins job prior to build. The input file that user is going to upload is not necessarily a text or a property file.
I am specifically trying to get the user to "select" their desired file - browse to their file ; the user should not pass the file's path.
Thanks
Use the File Parameter:
File parameter allows a build to accept a file, to be submitted by the user when scheduling a new build. The file will be placed inside the workspace at the known location after the check-out/update is done, so that your build scripts can use this file.
If you need to verify the file has a certain extension, you would have to do that with a script as part of your job, and fail the job is extension/content-type does not match what you need.
This is kind of annoying to handle when you don't know what the file name will be or need to change its name before it reaches its destination. You kind of need to perform a hack. This is how I do it:
Use the "File parameter" parameter to upload your file
Use the OS-specific script to rename the file from whatever you named your File Parameter to whatever you want it to be, e.g., if my File Parameter had the File location value of file_name instead of an actual relative file-path, I'd then do something like this for say, Windows inside a Build-Step for "Execute Windows Batch Command":
move .\file_name .\%file_name%
And then just use ArtifactDeployer to copy everything there to your desired location.
ps: this won't remove digital signatures, so the move-operation should be considered mostly safe.
The use of the Jenkins File Parameter will not work for Jenkins pipelines. It's ridiculous that they don't disable that kind of build parameter for pipelines. It's even more ridiculous that they don't at the very least, identify this SEVERE limitation in the help documentation for that parameter.
It would have saved me a couple hours trying to figure out why it would not work in my pipeline.
Refer to this feature request for more details: https://issues.jenkins-ci.org/browse/JENKINS-27413
I have a parametrized Jenkins task. After the task execution I can see the parameters used for the execution.
I want to reuse parameters of any previous build of the task for following launches.
Brief googling did not help me.
Could you please advise me some solutions for my task?
Use the Rebuild Plugin -
It allows you to rerun any previous run of a job,
with the same parameters that were used in that run.
You can even change some of the parameters, if needed.
You could do the following:
Save the parameters to a file in properties file format (e.g. using some of the script execution builders)
Save this file to a known location within the job
Use the Parameterized Trigger Plugin to reuse the parameters
I have about 20 jobs using common parameters (user, password), and sometimes the password expires... So I have to change it on all jobs, which is really time consuming (and error prone, I may forget one).
I thought about:
using a kind of magic property file if that exists to have directly lines like KEY, VALUE added into job parameters
adding the same kind of KEY, VALUE pair directly inside build.xml, but where ? And it's really ugly... Maybe with a dedicated XML embedded into the build.xml ?
calling a slave job that would (how ?) push up to the parent one the desired values...
As you can see I'm only starting in Hudson/Jenkins (I'm using Jenkins 1.424.2.2), thanks for your help !
EDIT: I'm not admin of the Jenkins instance, so I cannot have access to global properties...
Go to your Jenkins home and navigate :
Manage Jenkins >
Configure System >
Global properties >
Environment variables > ....
I can think of two approaches:
Use Global properties, found under Manage Jenkins -> Configure system. Here you can define environment variables that should be available to all jobs.
Write a small script that downloads, modifies and posts the job config: http://[jenkinshost]/job/[jobname]/config.xml.
You can read about the api capabilities under http://[jenkinshost]/job/[jobname]/api, here is what it says about reading and changing config.xml:
Fetch/Update config.xml
To programmatically obtain config.xml, hit
[http://[jenkinshost]/job/[jobname]/config.xml]. You can also POST
an updated config.xml to the same URL to programmatically update the
configuration of a job.
I eventually succeeded by:
keeping encrypted credentials in a web page
retrieving them in Hudson thanks to a shell script (wget), decrypt them, and creating a build.properties file in the workspace with lines name=value (in my case ssh.password=...)
This works, because Ant build steps detect this file and pass the variables inside into their context. Thanks to that I could centralize my credentials.