Different promotion behaviour causes failures if running on master

5 views
Skip to first unread message

John Mellor

unread,
Jul 28, 2015, 11:49:59 AM7/28/15
to jenkins...@googlegroups.com
If Jenkins selects a slave to run a promotion on, it assumes that the working directory is /var/lib/Jenkins/workspace/<project> and correctly copies the artifact tree from /var/lib/jenkins/jobs/<project>/builds/<build-number>/archive contents.
However, if it selects the master node to run on, it assumes a working directory of /var/lib/jenkins/jobs/<project>/workspace and fails to copy the artifact tree from ../builds/<build-number>/archive contents for unknown reasons.  I can find no log to explain why it malfunctions like this.
 
The workaround that I’m using right now is to disable the selection of the master node for all builds, which is not desirable from a cluster performance or cluster size point of view for the long term.  I’m using a pretty recent Jenkins 1.620 on Ubuntu 12, and do not see any fixes related to this in the 2 newer minor Jenkins releases.
 
How do I get a promotion running on master to work?  Is this a known bug, and the master node is always going to fail a promotion?
Reply all
Reply to author
Forward
0 new messages