Having tasks in pending status before reaching num of worker limit - SonarQube Enterprise 7.7

SonarQube Server / Community Build
1

Hi there,

We are new to SonarQube Enterprise and are experiencing a problem with many tasks in the pending status while the number of in-progress tasks is lower than the number of concurrent workers configured.

The number of Worker Count is 6, the number of in-progress tasks is less than 3, whilst the Pending tasks can sometimes higher than 10. The installation is on an EC2 with 8 vCPU and 32gb memory.

Must-share information (formatted with [Markdown]

  • which versions are you using (SonarQube, Scanner, Plugin, and any relevant extension)

    1. SonarQube 7.7.0.23042 Enterprise Edition
    2. Scanner - sonar-scanner-protocol-7.7.jar
    3. Plug-ins:
      SonarJava: 5.12.1.17771
      SonarKotlin: 1.5.0.315
      SecurityJavaFrontend: 7.7.0.4011

  • what are you trying to achieve
    Increase the number of concurrent workers and reduce/shorten the number and time a task being in the pending status.

  • what have you tried so far to achieve this?
    I have upgraded the license, upgraded the server specs, followed this guide to tweak the configurations https://docs.sonarqube.org/latest/instance-administration/compute-engine-performance/ . I’ve run out of ideas, please help.

Cheers,
Ryan

2

Hi Ryan,

Take a look, please at your Administration > System page. Expand the Comput Engine Tasks and look at the worker counts. What do you see there? Here’s what I see in one of our systems:

Selection_452

 
Ann

3

Hi Ann,

Thank you for your reply, the Worker Count is 6. I didn’t set it to 10 since we only have an EC2 instance with 8 vCPUs.

image
image.png633×674 60.6 KB

Cheers,
Ryan

6

Hello Ryan,

Is it possible that your multiple tasks that are pending are related to the same SonarQube project?

There is a limitation right now in situations where multiple branches/pull requests of the same project are being analyzed. In these instances, only one task will go through at a time (branches interact with each other to some extent with regard to issue tracking)

This is a situation we want to improve, and there’s a feature out there (MMF-1543) you can take a look at.

Let me know if this sounds like what’s going on with your SQ instance.

Colin

9

Hi Colin,

Thank you for your reply. The issue is very likely to be caused by the problem that you’ve mentioned. I will run some tests to confirm it.

Cheers,
Ryan