dependencies / requirements for the Jenkins vSphere Cloud plugin?

153 views
Skip to first unread message

Roger Moore

unread,
Nov 13, 2015, 2:50:28 PM11/13/15
to jenkins...@googlegroups.com

I am having problems with achieving the documented behavior of the vSphere Cloud plugin. I have read the description for the vSphere Cloud plugin ( https://wiki.jenkins-ci.org/display/JENKINS/vSphere+Cloud+Plugin) and followed the instructions on the page. And, I’ve looked through the jenkinsci-users group topics without finding the info described below.

 

In this situation, the node configuration matches that specified in the plugin instructions. However, I haven’t been successful getting Jenkins to start/stop/restart the Jenkins node as described therein. Previously though, I have successfully created a job that will stop, update the RAM, and then restart a VM using the plugin. It just won’t start it when the VM is supposed to be a Jenkins node. There are no log messages of interest.

 

Some system configuration details:

 

Master is CentOS 7, java 1.7.0_91, current Jenkins LTS, latest version of VMware tools.

Agent is Windows Server 2012R2, Java 1.8.0_54, and latest version of the VMware tools.

vCenter machine is Windows Server 2012R2 and vCenter 6.0 (no updates). Java is not installed independently.

 

Some of the causes for concern include:

-        It does not create the Jenkins installation directory on the node.

-        The node was originally figured to suspend when nothing was going on. It never wakes up, so I changed that in Jenkins to never power off. But it continues to suspend. It took a reboot of both the node and master before it would quit being suspended for lack of use.

-        In the Jenkins GUI in the browser, there is no ‘start Jenkins on node’  button like there is for non-vSphere plugin controlled nodes to manually start Jenkins on the node.

-        I tried running a job that uses the node when the node is powered off, suspended, and powered on. The Jenkins job never makes it out of the queue because the node never become active.

-        I know communications between the node and master used to work because the node had been configured as a node (then deleted) before being readded as the node through the plugin.

 

Therefore, I am wondering if there are some inherent / unspoken requirements that I’m missing. Does anyone know  where I can find information about that?

 

For example:

-        What user will own the agent’s Jenkin directory structure and run the associated java process? The only user specified in the plugin is to access the vCenter machine. But, there are 3 possibilities: a system account (on Windows), the plugin specified account to access the vCenter machine, and the user the Jenkins daemon (we’re on CentOS) on the master is running as. I don’t see that documented anywhere and the answer will require additional configuration. It seems like the Jenkins daemon account is the most likely, but in the current environment, that user does not exist on Windows because we had been using a specific Windows account for that.

 

Any information you have would be appreciated.

 

Reply all
Reply to author
Forward
0 new messages