Queued build timing is wrong

[felipecrs]

Jenkins and plugins versions report

Jenkins: 2.516.2
OS: Linux - 6.6.87.2-microsoft-standard-WSL2
Java: 21.0.8 - Eclipse Adoptium (OpenJDK 64-Bit Server VM)
---
pipeline-graph-view:628.va_6f6a_1d12848

What Operating System are you using (both controller, and any agents involved in the problem)?

Controller running as Docker, agents running through Kubernetes plugin.

Reproduction steps

1. Trigger a build

2. Check the queued time in the build page

   

3. Check the queued time in Pipeline Overview. Notice it is different.

   

4. You can also check in the Timings page

   

Expected Results

Should show the same timing as shown in the build page.

Actual Results

Seems to be showing a subset of it, not including subtasks.

In a real-world scenario here, different from these screenshots, Pipeline Overview shows that the build was queued for 18s while in reality it was queued for 13m, which is very misleading.

Anything else?

No response

Are you interested in contributing a fix?

No response

[KalleOlaviNiemitalo]

The "Timings" page is provided by the Metrics plugin. Coincidentally, I was going to uninstall the Metrics plugin from a Jenkins controller today because the timings were wrong and not useful, but couldn't because Pipeline Graph View depends on it:

pipeline-graph-view-plugin/pom.xml

Lines 81 to 84 in a6f6a1d

	<dependency>
	<groupId>org.jenkins-ci.plugins</groupId>
	<artifactId>metrics</artifactId>
	</dependency>

pipeline-graph-view-plugin/src/main/java/io/jenkins/plugins/pipelinegraphview/cards/items/TimingRunDetailsItems.java

Line 11 in a6f6a1d

import jenkins.metrics.impl.TimeInQueueAction;

This issue may need to be refiled in Jira for the metrics-plugin component. There are already JENKINS-54689 and JENKINS-71353 but I don't think they are quite the same thing as this.

What kind of agent directives are you using? In my case, I have agent none at the pipeline level, no agent directive in the first stage, and then a different agent in the second stage; so most of the the queueing time is spent between the first and second stage, and I think the Metrics plugin doesn't measure that.

[felipecrs]

What kind of agent directives are you using? In my case, I have agent none at the pipeline level, no agent directive in the first stage, and then a different agent in the second stage; so most of the the queueing time is spent between the first and second stage, and I think the Metrics plugin doesn't measure that.

pipeline {
  agent any
}

https://github.com/felipecrs/docker-images/tree/master/jenkins-agent-dind#pod-template-with-the-kubernetes-plugin-on-jenkins

[felipecrs]

This issue may need to be refiled in Jira for the metrics-plugin component. There are already JENKINS-54689 and JENKINS-71353 but I don't think they are quite the same thing as this.

I don't think so. In my case the timings provided by Metrics plugin are accurate, and Pipeline Overview is simply taking the data from the wrong location.

[KalleOlaviNiemitalo]

You wrote that "Pipeline Overview shows that the build was queued for 18s while in reality it was queued for 13m" but the Timings screenshot does not show the 13m queue duration, either.

[felipecrs]

SORRY!

I just realized now that I made this claim very confusing. I changed that sentence to:

In a real-world scenario here, different from these screenshots, Pipeline Overview shows that the build was queued for 18s while in reality it was queued for 13m, which is very misleading.

[felipecrs]

Here is the real-world scenario I referred to:

11:06:29  Waiting for next available executor on ‘dind’
11:18:32  Agent dind-xpt0d is provisioned from template dind