We have a docker image say xyz:latest which needs to be pushed to my docker registry xxx.com at repo "docker-local" via Jenkins Pipeline
Things were working fine earlier but after updating the artifactory plugins over Jenkins, we are facing the issue with
"INFO: The search failed with "Failed to retrieve repository configuration 'docker-local': HTTP/1.1 400 Bad Request"."
With reference to the code given at (https://www.jfrog.com/confluence/display/JFROG/Setting+Up+Docker+Build-info), more particularly the step 3 part of it, can someone please explain out what does it exactly mean??
// Step 3: Push the image to Artifactory.
// Make sure that <artifactoryDockerRegistry> is configured to reference <targetRepo> Artifactory repository. In cas
def buildInfo = rtDocker.push '<artifactoryDockerRegistry>/hello-world:latest', '<targetRepo>'.
What my values currently for them are:
artifactoryDockerRegistry: "xxx.com/docker-local/xyz:latest"
targetRepo: "docker-local"
Somehow I land-up with the error as mentioned in the comments section of Step3 above.
Can someone please help me with a generic example as to what should the values be??
Related
I have signed up for one of the public docker registries, so I've been given a username and password. I'm writing a Jenkins job which pulls an image from this repository, so I'm using the following command in my Jenkins pipeline
docker.withRegistry('https://registry.example.com', 'credentials-id')
However I don't know what I should put in as the credentials-id? How can I get it?
This credentials-id Item is provided by the Jenkins credentials Plugin. It is documented here, e.g. https://jenkins.io/doc/book/using/using-credentials/
I have a JSP website. I am building DevOps pipeline. I am looking for help to integrate Jenkins with the Docker.
I already have docker file which does task of Deploying war file to the tomcat server.
(Command1)
Through the command line I can run the docker file and create an image.
I can run created image as a service and able to browse the website.
(Command2)
I want to do these two steps in Jenkins. I need your help to integrate these two commands in Jenkins, so that I need not to run these two commands manually one after other.
I think that you can use the "Docker Pipeline Plugin" for that.
For the first command, you can have a stage that runs:
myImage = docker.build("my-image:my-tag")
If you need you can have another stage where you can run some tests inside the image with:
myImage.inside {
sh './run-test.sh'
}
Finally, you can push the image to the repository to your repository with:
docker.withRegistry('https://your-registry.com', 'credentials_id') { //use a second parameter if you repository requires authentication
myImage.push('new_tag') //You can push it with a new tag
}
Please note that if you wanna use the docker.* methods in a declarative pipeline you must do it inside a script step or in a function.
(More info in the plugin's user guide)
For the second command, you only have to update the running image in the server. For doing that you have a lot of options (docker service update if you're using Docker Swarm, for example) and I think that part is outside of the scope of this post.
I'm using a Docker Registry on Artifactory. I'm able to pull/push images using docker commands. Now I try to push an image using a Jenkins pipline.
The image is called registry-url/docker/image:latest.
I have a docker repository on Artifactory which is called docker. (I'm able to pull and push to this repo using docker commands).
This stage describes my Artifactory configuration:
...
stage('Deploy Docker image'){
steps {
script {
def server = Artifactory.server 'xxx'
def rtDocker = Artifactory.docker server: server
def buildInfo = rtDocker.push('registry-url/image:latest', 'docker')
//also tried:
//def buildInfo = rtDocker.push('registry-url/docker/image:latest', 'docker')
//the above results in registry/docker/docker/image..
server.publishBuildInfo buildInfo
}
}
}
...
When I use different paths I face the manifest.json error which is probably normal.
I'm able to download the manifest.json manually on: https://registry-url/artifactory/docker/image/latest/manifest.json.
I'm using a pretty new version of Docker on Jenkins:
Docker version 18.01.0-ce, build 03596f51b1
So far so good. But When I run the pipeline I receive the following error in Jenkins (it takes 50 seconds):
Pushing image: registry-url/image:latest
...
com.github.dockerjava.api.exception.DockerClientException: Could not push image: unknown: Not Found
at com.github.dockerjava.core.command.PushImageResultCallback.awaitSuccess(PushImageResultCallback.java:49)
at org.jfrog.hudson.pipeline.docker.utils.DockerUtils.pushImage(DockerUtils.java:60)
at org.jfrog.hudson.pipeline.docker.utils.DockerAgentUtils$3.call(DockerAgentUtils.java:213)
at org.jfrog.hudson.pipeline.docker.utils.DockerAgentUtils$3.call(DockerAgentUtils.java:205)
at hudson.remoting.UserRequest.perform(UserRequest.java:153)
at hudson.remoting.UserRequest.perform(UserRequest.java:50)
...
In Artifactory logs I see:
2018-04-25 14:24:26,663 [http-nio-8081-exec-xx] [ERROR] (o.a.a.d.r.DockerResource:153) - Unsupported docker v2 repository request for 'image'
2018-04-25 14:24:46,684 [http-nio-8081-exec-xx] [ERROR] (o.a.a.d.r.DockerResource:153) - Unsupported docker v2 repository request for 'image'
2018-04-25 14:24:46,689 [http-nio-8081-exec-xx] [ERROR] (o.a.a.d.r.DockerResource:153) - Unsupported docker v2 repository request for 'image'
2018-04-25 14:24:46,702 [http-nio-8081-exec-xx] [ERROR] (o.a.a.d.r.DockerResource:153) - Unsupported docker v2 repository request for 'image'
What am I missing or doing wrong?
EDIT:
Based on this issue I went back to my initial idea:
def buildInfo = rtDocker.push('registry-url/docker/image:latest', 'docker')
I tried the build again. Error:
Could not find manifest.json in Artifactory in the following path: https://registry-url/artifactory/docker/docker/image/latest/manifest.json
Two times 'docker' in the path and it seems not to work. BUT when I check in Artifactory the image is there... I can also pull the image. It seems to be fine but still the jenkins build is failing.
Artifactory Plugin: 2.15.1
Artifactory Version: 5.10.3
Is this really a bug which will be fixed soon?
Artifcatory can be configured as a docker registry either with or without a reverse proxy.
It looks like your Artifactory is not configured using a reverse-proxy (proxy-less configuration). You can read more about the configuration options here.
Version 2.16.1 of the Jenkins Artifactory Plugin added support for proxy-less configuration. Upgrading your Artifactory Plugin should resolve your issue.
Try following this example. Here we have Jenkins pipeline to pull/push docker image to/from Artifactory: https://github.com/jfrogtraining/kubernetes_example/blob/master/docker-app/Jenkinsfile#L43
I try to get the list of published Artifacts from my deploy onto my Artifactory.
I tried to do so over the BuildInfoAccessor, but the current version is lacking the getDeployedArtifacts() function.
I even tried to read the jenkins build.log object, but it somehow misses the prints from artifactory-plugin on which artifacts are deployed.
Can someone give me a hint on where to look at or an example ?
Am not sure is there any better way to print the BuildInfo from the artifactory-jenkins plugin itself. You can get the published info of the jenkins build via artifactory rest api.
Artifactory Rest Api
You can get the build number from the jenkins environment variable ${BUILD_NUMBER} and make a http get call via sh curl/other suitable step if any in your pipeline script.
sh "curl http://artifactory.org.net/api/build/my-build/${BUILD_NUMBER}"
Make use of withCredential step to pass username/password.
Caution: I have just made answers from my theoretical knowledge.
I have GitLab Community Edition 8.15.2 successfully trigger pipeline projects in Jenkins 2.32.1 using a webhook. I want the gitlab push to trigger a build with parameters but the parameter value is null when it comes through so the build fails.
The gitlab webhook looks like:
http://jenkins.server:8080/project/project-a/buildWithParameters?MYPARAM=foo
In my pipeline project I echo the parameter value out with
echo "MYPARAM: ${MYPARAM}"
and it's not set to anything. Any ideas on where I've gone wrong?
UPDATE
The actual code I'm using in the pipeline is:
node {
try {
echo "VM_HOST: ${VM_HOST}"
echo "VM_NAME: ${VM_NAME}"
stage('checkout') {
deleteDir()
git 'http://git-server/project/automated-build.git'
}
stage('build') {
bat 'powershell -nologo -file Remove-MyVM.ps1 -VMHostName %VM_HOST% -VMName "%VM_NAME%" -Verbose'
}
...
}
}
The parameter VM_HOST has a default value but VM_NAME doesn't. In my Console output in Jenkins I can see:
[Pipeline] echo
VM_HOST: HyperVHost
[Pipeline] echo
VM_NAME:
I have been struggling with this for weeks. I had it working once, but I couldn't get it to work again, untill today. And the solution was mindblowingly obvious ofcourse...
Automatically for each pipeline job I ticked the following box:
Build when a change is pushed to GitLab. GitLab CI Service URL:
http://jenkins.dev:8080/project/MyProject
Then from GitLab I used the webhook to trigger the above.
Like you I tried to add /buildWithParameters and tried many other things that didn't work.
The problem was, I ticked the wrong checkbox!
Since I trigger the build from a GitLab webhook, the above checkbox (build when a...) does not have to be checked at all.
What needs to be checked is:
Trigger builds remotely (e.g., from scripts)
That checkbox provides you with a new URL:
Use the following URL to trigger build remotely:
JENKINS_URL/job/MyProject/build?token=TOKEN_NAME or
/buildWithParameters?token=TOKEN_NAME
Like all the documentation I came along states and as you can see, the URL now no longer starts with /project, but with /job instead!
So tick that box and change your URL accordingly:
http://jenkins.server:8080/**job**/project-a/buildWithParameters?token=TOKEN_NAME&MYPARAM=foo
Least I want to mention the token:
In the GitLab webhook there is a seperate field for "token", which states:
Use this token to validate received payloads. It will be sent with the request in the X-Gitlab-Token HTTP header.
So, the token provided there will be sent along the request as a HTTP header.
This is the token which can be provided globally in the Jenkins setup.
The token you must provide in the Jenkins job when ticking the box Use the following URL to trigger build remotely must be send in the URL as GET parameter, just like the example shows.
Final note: personally I have never got this working completely, because I don't get the Jenkins CSRF protection off my back. Disabling it gives me another error. However, hopefully the above does fix the problem for you and others.
GitLab plugin does not allow you to pass arbitrary parameters. In their project there is an open issue for it that deserves to be upvoted.
My convoluted solution was to use the desired values for the push trigger as the default parameters of the job. Then I used the Parameterized Scheduler plugin to use other values in the scheduled executions.
The problem is that I got a bad usability for the job when it was manually run, since the default parameters were appropriate for the push hook.
I found the solution here https://www.jittagornp.me/blog/jenkins-gitlab-webhook/
I verified it with Jenkins 2.263.1 and GitLab Community Edition 13.6.1
Your webhook url will look like
https://hunter:11a403302a4f01b9b4975c0ac27441a5cc#jenkinsservername.com/job/yourjenkinsproject/buildWithParameters?token=Aju9ryHUu6t7W8wLSeCWtY2bWjzQduYNPyY7B3gs&yourparam=yourvalue
"hunter" ist your username in Jenkins.
The following is the Jenkins API Token you have to create in your Jenkins User Managment independent of the project.
The last Token is the one you specify in the jenkins project options under "Trigger builds remotely (e.g., from scripts)"
The last thing is to add your Parameter and value to the url with ¶m=value