-
Improvement
-
Resolution: Unresolved
-
Major
-
None
-
Jenkins v1.565.3
performance-plugin v1.13
We have a nightly Jenkins job that uses a slave to run JMeter and publish the results using the performance plugin.
The slave that we use is a VM and is reset every night.
In the Jenkins configuration for the node:
Availability: 'Take this slave on-line when in demand and off-line when idle' In demand delay: 1 (minute) Idle delay: 60 (minutes)
It is observed that when the job is completed the graph is shown (as expected). However, after the idle delay is elapsed, the slave is disconnected and goes offline and the performance graph disappears.
Only by bring the node online again will the graph be available.
I am not sure if this is a unique problem that affect all plugins. Could something be done to 'cache' the information on the server, similar to how the artifact, trends, etc are available even if the slave is offline?
[JENKINS-27245] Performance graph does not appear if the slave node is offline
suresh nallamilli
added a comment - won't be able to contribute in next couple of weeks , if some one else can pick it up.. thats great
suresh nallamilli
added a comment - won't be able to contribute in next couple of weeks , if some one else can pick it up.. thats great 
Paul Cosma
added a comment - I have the same issue for the jobs that are running tests using docker container slaves (with docker plugin) as it is a short-lived slave.
Paul Cosma
added a comment - manolo
I have the same issue for the jobs that are running tests using docker container slaves (with docker plugin) as it is a short-lived slave. 
Philippe Mouawad
added a comment - Cannot reproduce as of 3.15 version.
Slave offline, still graphs are displayed.
Closing
Philippe Mouawad
added a comment - Cannot reproduce as of 3.15 version.
Slave offline, still graphs are displayed.
Closing I can still reproduce it on
Jenkins ver. 2.150.3, Performance Plugin vers. 3.15
using ephemeral docker build slaves.
Docker plugin
https://plugins.jenkins.io/docker-plugin
jenkins ssh slave
https://hub.docker.com/r/jenkins/ssh-slave
Please inspect attachments.
The graphs will work only if you perform the workaround suggested by @suresh nallamilli
Paul Cosma
added a comment - - edited I can still reproduce it on
Jenkins ver. 2.150.3, Performance Plugin vers. 3.15
using ephemeral docker build slaves.
Docker plugin
https://plugins.jenkins.io/docker-plugin
jenkins ssh slave
https://hub.docker.com/r/jenkins/ssh-slave
Please inspect attachments.
The graphs will work only if you perform the workaround suggested by @suresh nallamilli 
I too faced same issue today. During my debugging i found that this will happen if jobs are configured to run on short-lived slaves (Cloud instances ) .
This is due to below piece of code in PerformanceProjectAction.java
this.project.getSomeBuildWithWorkspace().getRootDir()
getSomeBuildWithWorkspace checks for a live work space for any build and this causes issues if none of the previous builds have live workspace.
Temporary and ugly work-around is to run the job on jenkins master once.
I will try to submit pull request for this.