问题
Is it possible that for a given Build Pipeline job (which has downstream jobs either in the build or post build action as "Trigger build on other projects"), I can get a Tree listing view showing which Pipeline job# N called, what child jobs in the calling order (sequential or parallel) with child build# for that pipeline run build#.
For ex: If my pipeline job has this view: then,
I'm expecting to get a listing of the top run similar to (in case I just put in simple text format):
vac-3.0-src:52 called: vac-3.0-unit-test-main:37
vac-3.0-unit-test-main:37 called: vac-3.0-unit-testA:36
vac-3.0-unit-test-main:37 called: vac-3.0-unit-testB:36
vac-3.0-unit-test-main:37 called: vac-3.0-unit-testC:35
vac-3.0-unit-test-main:37 called: vac-3.0-unit-testD:35
vac-3.0-unit-test-main:37 called: vac-3.0-unit-testReporting:35
vac-3.0-unit-testReporting:35 called: vac-3.0-integration-test-main:28
vac-3.0-integration-test-main:28 called: vac-3.0-integration-testA:27
vac-3.0-integration-test-main:28 called: vac-3.0-integration-testB:27
vac-3.0-integration-testB:27 called: vac-3.0-acceptance-test:25
vac-3.0-acceptance-test:25 called: vac-3.0-configure-something:24
vac-3.0-configure-something:24 called: vac-3.0-perform-someaction:23
vac-3.0-perform-someaction:23 called: vac-3.0-preview-step:22
vac-3.0-preview-step:22 called: vac-3.0-deb-delivery-job:27
vac-3.0-preview-step:22 called: vac-3.0-rpm-el6:23
vac-3.0-preview-step:22 called: vac-3.0-vagrant-provision:20
vac-3.0-preview-step:22 called: vac-3.0-vagrant-run:21
vac-3.0-vagrant-run:21 called: vac-3.0-demo:10
OR this information can be presented in a more robust structural manner i.e. it can be JSON blob where a parent job has structure which will have all jobs that it called (parallel/sequence) in the pipeline run / given order.
I tried the main job's URL (via curl) with Jenkins API i.e. /api/xml or /api/json?pretty=true&depth=10 or more but it doesn't give me the information that I'm looking for (related to a given pipeline run).
This information is visually available on the pipeline view (as per the image) and some information about subprojects is available on a given Jenkins job's dashboard (which was part of the pipeline) but the order is not there.
I'll appreciate if you have tried to solve this and have any solution to get this data. The reason for this effort is to find metrics horizontally for a given pipeline run (Rather than vertically for each individual job which are part of pipeline as I already have the vertical / individual job metrics for Total time, build#, result etc) but how I can relate each individual job's metrics for a given pipeline run, is what I'm trying to get.
If the above image example is big enough, we can refer a smaller run image snapshot here:
回答1:
I see one possible solution, not sure if that's helpful but surely it's an attempt.
Algorithm Steps:
==============
1) Maintain a Direct Parent-Child file (i.e. JobA:JobB, JobA:JobC, JobA:JobC, JobC:JobD, ....) i.e. this file will tell that for each JobX, what's are the direct Sub-child/downstream jobs of that. Via Jenkins Groovy script, this can be easily generated/available. PS: You can add more columns to this file i.e. JobA:JobB:Build:Sequential or JobA:JobB:Test:Parallel to get even better horizontal metrics for calculating turnaround time / per given step (build, test, deploy, etc) and whether a parent job called the child job in a sequence or in parallel with two or more jobs) and calculate the metrics accordingly.
2) Inside "Build Pipeline View" Configure (Settings), set the no. of jobs to be displayed as 1.
PS: You can set this to whatever 5, 10, or more if you want to capture a given pipeline build# of that main pipeline job.
For testing purpose, I'm showing only 1 pipeline build run.
3) In Linux, use curl, get the "View Source" HTML page information on the build-pipeline-view's NAME (PS: This is NOT on the main pipeline job).
i.e. **not for jobA or xxvt-main or ** in this case, but use the View Name URL (which shows the whole pipeline). Let's assume the view name (via Build Pipeline View plugin) was created as "MyPipelineView"
ex: curl -s http://my-jenkins-server:8080/view/MyPipelineView/ > /tmp/9.txt
This will give you the HTML content.
Store this information in some file (Temporary). Let's assume I stored it in /tmp/9.txt
3) Run the following command to get the job's build#s. As per the second smaller pipeline image (in my post), the output of that will be:
grep -o "\"extId\":\"[a-zA-Z0-9_-][a-zA-Z0-9_-]*#[0-9][0-9]*\"" /tmp/9.txt
This will give you output like (use sed/cut to make it more cleaner):
"extId":"xxvt_main#157"
"extId":"xxvt_splunk_run_collect_operation#29"
"extId":"xxvt_splunk_run_process_operation#29"
"extId":"xxvt_splunk_update_date_restart_splunk#29"
"extId":"xxvt_splunk_get_jenkins_data#38"
"extId":"xxvt_splunk_get_clearquest_dr_data#47"
4) Now you have the above output for a given pipeline run, using the Parent-Child (direct relationship) file (which we generated in bullet 1), we can use that to create our final Build Pipeline Tree file i.e.
xxvt_main#157 called: xxvt_splunk_get_jenkins_data#38
xxvt_main#157 called: xxvt_splunk_get_clearquest_dr_data#47
xxvt_main#157 called: xxvt_splunk_run_collect_operation#29
xxvt_splunk_run_collect_operation#29 called: xxvt_splunk_run_process_operation#29
xxvt_splunk_run_process_operation#29 called: xxvt_splunk_update_date_restart_splunk#29
5) Upon knowing the a given run related job-name and its build#, we can use Jenkins's api/json?pretty=true&depth=1 or 2 or 3 carefully, to fetch fields we want to fetch for metrics and finally create/come up with a .csv file in whatever format you like, which will have metrics for a given pipeline run - HORIZONTALLY.
来源：https://stackoverflow.com/questions/50071652/jenkins-build-pipeline-job-jobs-listing-in-tree-layout-ordered-listing