nrayapati we'd be glad to look into this. Could you please attach a HAR file to this ticket replicating this issue? If you wish to keep the contents of the har file confidential, please email me at jdumay@cloudbees.com with "JENKINS-47480" as the subject and the HAR attached.

nrayapati if you have a simplified version of your Jenkinsfile that can reproduce this problem that would be greatly appreciated too.

We've also just hit this.  I am so far unsuccessful in reproducing it with a simplified Jenkinsfile.  We have other pipelines running in the same jenkins instance which haven't hit this problem yet, so its a bit wierd.

Also, this only happens on our master branch at the moment in that particular pipeline.  Other branches using the same Jenkinsfile seem fine.

Im also having this problem, it's on a job that has many many steps.  When you start running the build, it displays everything, until it gets most of the way through, then it starts throwing this error.

garrettmarone morriste would either of you be able to share your Jenkinsfile? You can email it to me at jdumay@cloudbees.com with the subject "JENKINS-47480" if you wish to keep it confidential. A HAR file of the session would be incredibly useful too.

Just for more information, the following comes out in the log:

Oct 20, 2017 3:42:06 PM io.jenkins.blueocean.rest.impl.pipeline.PipelineNodeGraphVisitor parallelStart
SEVERE: nestedBranches size: 1 not equal to parallelBranchEndNodes: 2
Oct 20, 2017 3:42:06 PM io.jenkins.blueocean.rest.impl.pipeline.PipelineNodeGraphVisitor parallelStart
SEVERE: nestedBranches size: 3 not equal to parallelBranchEndNodes: 4
Oct 20, 2017 3:42:06 PM io.jenkins.blueocean.rest.impl.pipeline.PipelineNodeGraphVisitor parallelStart
SEVERE: nestedBranches size: 16 not equal to parallelBranchEndNodes: 17

I've sent you through a har file too, thanks!

jamesdumay / cliffmeyers Sorry for the late reply, but I noticed that it is happening when we have a parallel step inside another parallel. Here is the sample example to reproduce. hope this is helpful.

def stepsForParallel = [:]
for(int i =0; i < 10; i++) {
stepsForParallel[i] = { 
def stepsForParallelInternal = [:]
for(int j =0; j < 10; j++) {
stepsForParallelInternal[j] = { echo j.toString() }
}
parallel stepsForParallelInternal
}
}
parallel stepsForParallel
stepsForParallel = [:]
for(int i =0; i < 20; i++) {
stepsForParallel[i] = { echo i.toString() }
}
parallel stepsForParallel
stepsForParallel = [:]
for(int i =0; i < 20; i++) {
stepsForParallel[i] = { echo i.toString() }
}
parallel stepsForParallel
stepsForParallel = [:]
for(int i =0; i < 20; i++) {
stepsForParallel[i] = { echo i.toString() }
}
parallel stepsForParallel
stepsForParallel = [:]
for(int i =0; i < 50; i++) {
stepsForParallel[i] = { echo i.toString() }
}
parallel stepsForParallel

We are also seeing this with the latest (public) version of Jenkins, and likewise didn't have it before. It is definitely something to do with parallel. I'm fairly sure we don't have parallel within parallel, but we do like the above generate the parallel statements from maps. I wonder if the issue lies there.

In further testing, again against the 1.3.2 docker image, this works:

def stepsForParallel = [:]
for(int i = 0; i < 98; i++) {
  stepsForParallel[i] = { echo i.toString() }
}
parallel stepsForParallel

stepsForParallel = [:]
for(int i = 0; i < 2; i++) {
  stepsForParallel[i] = { echo i.toString() }
}
parallel stepsForParallel

and this fails:

def stepsForParallel = [:]
for(int i = 0; i < 99; i++) {
  stepsForParallel[i] = { echo i.toString() }
}
parallel stepsForParallel

stepsForParallel = [:]
for(int i = 0; i < 2; i++) {
  stepsForParallel[i] = { echo i.toString() }
}
parallel stepsForParallel

The problem here is not nesting, but ending up on the border between two stages when the page ends.

The fix to request a much larger page size would prevent most people from running into this: https://github.com/jenkinsci/blueocean-plugin/commit/ce4a4b39c4ff2a2d979195da18430205957e7ca8. I've confirmed this resolves the issue in the above reproduction. Could that be released in 1.3.3 immediately cliffmeyers? This exact issue caused us 4 hours of chaos when we upgraded to 1.3.2 today, it really is a critical fix.

Ping to jamesdumay and vivek

Thanks benlangfeld. I have opened a PR https://github.com/jenkinsci/blueocean-plugin/pull/1556. Once it passes we can plan a 1.3.3 release.

Fixed in master and pending fix for 1.3.3 release

Fixed in release 1.3.3. Release is out but it might take 4-6 hours to get across mirrors.

I can confirm that this is fixed in 1.3.3 for our production use case.

Thanks vivek and benlangfeld Our pipeline are fixed too.

Testing Notes:

• For this particular issue, manual verification, using the pipelines kindly provided by benlangfeld, should suffice.
• Longer term:
  • Test cases with longer, more parallel, more complex Jenkinsfiles should be implemented to make sure that Blue Ocean properly renders them
  • A repo of pipelines which can be used to trigger past issues could be collected over time, and used to inform these test cases

Adding a link to the child ticket, which talks about exploring some more automated complex test cases.