导出 jenkins 作业的构建历史

Question

Can I export, in any conventional file format, a history of builds, with their time/date and success. And hopefully even promotion status.

Answer 1

You can make use of Jenkins rest api :

1. Start at : Traverse all jobs on your Jenkins Server using :
   http://JENKINS_URl/api/json?tree=jobs[name,url]
   This will give json response with all jobs with job name and job url.
2. Then for each job access its builds using api :
   http://JENKINS_URL/job/JOB_NAME/api/json?tree=allBuilds[number,url]
   This will give all the builds for job JOB_NAME and return json response with build number and build url.
3. Now Traverse each build using api :
   http://JENKINS_URL/job/JOB_NAME/BUILD_NUMBER/api/json
   This will give everything related to the build as json response. Like Build status, how build was triggered, time etc.

For automation, you can use bash, curl and jq to achieve this.

Have written small bash script to retrieve build status and timestamp for each job on Jenkins server :

#!/bin/bash
JENKINS_URL=<YOUR JENKINS URL HERE>
for job in `curl -sg "$JENKINS_URL/api/json?tree=jobs[name,url]" | jq '.jobs[].name' -r`; 
do 
    echo "Job Name : $job"
    echo -e "Build Number\tBuild Status\tTimestamp"
    for build in `curl -sg "$JENKINS_URL/job/$job/api/json?tree=allBuilds[number]" | jq '.allBuilds[].number' -r`; 
    do 
        curl -sg "$JENKINS_URL/job/$job/$build/api/json" | jq '(.number|tostring) + "\t\t" + .result + "\t\t" + (.timestamp|tostring)' -r
    done 
    echo "================"
done

Note : Above script assumes that Jenkins server does not have any authentication. For authentication, add below parameter to each curl call :
-u username:API_TOKEN
Where :
username:API_TOKEN with your username and password/API_Token

Similar way you can export all build history in any format you want.

Answer 2

Parvez' suggestion to use the REST API is perfectly fine.

However, the REST API is awkward to use if it does not directly provide the data you're looking for, leading to convoluted and multiple invocations of the REST API. This is slow and it makes you depend on stability of that API.

For anything but the most basic queries, I usually prefer to run a small groovy script that will extract the required data from Jenkins' internal structures. This is way faster, and often it's also more simple to use. Here's a small script that will fetch the data that you're looking for:

import jenkins.model.*
import hudson.plugins.promoted_builds.*
import groovy.json.JsonOutput

def job = Jenkins.instance.getItemByFullName( 'TESTJOB' )

def buildInfos = []
for ( build in job.getBuilds() ) {
  def promotionList = []
  for ( promotion in build.getAction(PromotedBuildAction.class).getPromotions() ) {
    promotionList += promotion.getName()
  }
  buildInfos += [
    result    : build.getResult().toString(),
      number    : build.getNumber(),
      time      : build.getTime().toString(),
      promotions: promotionList
  ]
}
println( JsonOutput.toJson( buildInfos ) )

The script will produce the result in JSON format, like this (prettified):

[
    {
        "number": 2, 
        "promotions": [
            "promotionA"
        ], 
        "result": "SUCCESS", 
        "time": "Thu Oct 18 11:50:37 EEST 2018"
    }, 
    {
        "number": 1, 
        "promotions": [], 
        "result": "SUCCESS", 
        "time": "Thu Oct 18 11:50:34 EEST 2018"
    }
]

You can run such a script via the Jenkins "Script Console" GUI, or via the REST API for running groovy scripts (sic). There's also a CLI interface command for doing that.