Post-Commit Hook to trigger automatic Jenkins Build

Question

I know that there are many similar postings, but I have not found a solution, and the advice and solutions presented in the other posts don't quite jive with what I'm seeing.

The scenario is pretty darn simple: I have a project in Eclipse, and when I check-in changes from that project to our Subversion server (i.e., VisualSVN Server 2.5.3), I want our Jenkins continuous integration server (i.e., Jenkins 1.546) to pick up this change and kick off a new build. I do not want to poll from Jenkins.

I've been mostly following the steps in this article. Here's my post-commit hook script:

repos   = WScript.Arguments.Item(0)
rev     = WScript.Arguments.Item(1)
svnlook = WScript.Arguments.Item(2)
jenkins = WScript.Arguments.Item(3)

Set objFSO = CreateObject("Scripting.FileSystemObject")
Set objFile = objFSO.CreateTextFile("C:\Program Files (x86)\VisualSVN Server\log.txt")

objFile.Writeline "repos=" & repos
objFile.Writeline "rev=" & rev
objFile.Writeline "svnlook=" & svnlook
objFile.Writeline "jenkins=" & jenkins

Set shell = WScript.CreateObject("WScript.Shell")

Set uuidExec = shell.Exec(svnlook & " uuid " & repos)
Do Until uuidExec.StdOut.AtEndOfStream
  uuid = uuidExec.StdOut.ReadLine()
Loop

objFile.Writeline "uuid=" & uuid

Set changedExec = shell.Exec(svnlook & " changed --revision " & rev & " " & repos)
Do Until changedExec.StdOut.AtEndOfStream
  changed = changed + changedExec.StdOut.ReadLine() + Chr(10)
Loop
objFile.Writeline "changed=" & changed

url = jenkins + "crumbIssuer/api/xml?xpath=concat(//crumbRequestField,"":"",//crumb)"
Set http = CreateObject("Microsoft.XMLHTTP")
http.open "GET", url, False
http.setRequestHeader "Content-Type", "text/plain;charset=UTF-8"
http.send
crumb = null

objFile.Writeline "rev url=" & url
objFile.Writeline "http.status=" & http.status
objFile.Writeline "http.responseText=" & http.responseText

if http.status = 200 then
  crumb = split(http.responseText,":")
end if

url = jenkins + "subversion/" + uuid + "/notifyCommit?rev=" + rev + "&token=pinkfloyd65"
objFile.Writeline "url=" & url

if not isnull(crumb) then 
    objFile.Writeline "crumb(0)=" & crumb(0)
    objFile.Writeline "crumb(1)=" & crumb(1)
end if

if isnull(crumb) then 
    objFile.Writeline "crumb=null"
end if

Set http = CreateObject("Microsoft.XMLHTTP")
http.open "POST", url, False
http.setRequestHeader "Content-Type", "text/plain;charset=UTF-8"
if not isnull(crumb) then 
  http.setRequestHeader crumb(0),crumb(1)
  http.send changed
  if http.status <> 200 then
    objFile.Writeline "Error. HTTP Status: " & http.status & ". Body: " & http.responseText
  end if

  if http.status = 200 then
    objFile.Writeline "HTTP Status: " & http.status & ".\n Body: " & http.responseText
  end if
end if

The issue is that, although the POST command above ends up getting a 200 response back, the job never kicks off. Nothing ends up happening. Alright, so let's check the Jenkins job configuration; maybe I'm missing a setting or something. Well, under the Build Triggers section, I've checked the option for "Trigger builds remotely (e.g., from scripts)" and I've supplied an authentication token as well. But, the directions underneath that section look different from what I've been doing:

Use the following URL to trigger build remotely: JENKINS_URL/job/<job-name>/build?token=TOKEN_NAME or /buildWithParameters?token=TOKEN_NAME Optionally append &cause=Cause+Text to provide text that will be included in the recorded build cause.

So, it seems like there's a delta between the sets of instructions I'm seeing, and I'm not sure how to bridge that gap. It seems pretty obvious to follow the instructions on the Jenkins job configuration page, except that I don't know how I'd get the job name, rather than the UUID.

Another thing to note is my repository setup. Since the CI server is used by many groups and departments, I thought I'd be all smart and create a top-level repository to house just my department's projects. So, I have a setup something like:

VisualSVN Server  
  -- Repositories  
     -- Project_A  
     -- Project_B  
     -- <my-department>  
        -- DepartmentProject_A  
        -- DepartmentProject_B  

I'm wondering if the repository structure is adding to my issues here, but I feel like I should be able to find out which specific repository any changes came from. If that were true, then I could adjust my script to use the job name, rather than UUID, and then follow the explicit instructions seen on my CI server's configuration page. When I log the incoming repos variable in my vbs script, it points to the top-level department repository, rather than the project's child repository (i.e., D:\<visual-svn-repos>\<my-department> rather than D:\<visual-svn-repos>\<my-department>\DepartmentProject_B).

Any help would be greatly appreciated, thanks guys.

Answer 1

The article you linked says

Jobs on Jenkins need to be configured with the SCM polling option to benefit from this behavior. This is so that you can have some jobs that are never triggered by the post-commit hook (in the $REPOSITORY/hooks directory), such as release related tasks, by omitting the SCM polling option. The configured polling can have any schedule (probably infrequent like monthly or yearly). The net effect is as if polling happens out of their usual cycles.

and

For this to work, your Jenkins has to allow anonymous read access (specifically, "Job > Read" access) to the system. If access control to your Jenkins is more restrictive, you may need to specify the username and password, depending on how your authentication is configured.

Does you server meets this restrictions?