Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 47 Next »

Working page for information about

Current administrator: Gunnar Andersson  Please report any noticed problems immediately
We have had a few different secondary maintainers but right now that role is vacant - contact Gunnar if you want to be more involved.

Additional information (sub-pages)

How do I use it?

Q: Do I need a personal account?

A: You need a personal account if you want to edit/add jobs or to manually trigger the pipelines to run. 
The guest account
  (user: guest, password: genivigo) is sufficient to study the results of existing pipelines.    

If you want a personal account, please see:


1) To use, you do not need to install anything, just use a web browser and use the server with credentials as above.  Skip to the next section.

2) If you can provide a Go Agent machine (i.e. a "build slave" machine part of the shared build cloud), then refer to:

3) If you are interested in running your own test environment (both server and agent) locally, you might try easy-genivigo project (easy)
    Or modify the scripts and configuration used for the official GENIVI  go-server and go-agents (a bit more to do)

If you are an IT administrator and need to reinstall the site then instructions are at the end of Go Installation Instructions.

Organization (for developers and users)


You may want to watch or modify the queue of JIRA tickets.  If you have a JIRA login then it's best to look at the real JIRA task filter which shows Go-related tickets.

You can get a login here.  Alternatively, the Wiki assembls this report from JIRA right here in the Wiki.


Go Agents are what other systems might call "build slaves".  They do the actual compilation Jobs assigned by the Go Server.

List of Go Agents and maintainers
(This list needs maintenance - please notify the administrator if you see a mismatch)

Agent NameLocation, Company and Maintainer contactOperating system & setupComments


GENIVI Data Center
(Go issues : Gunnar Andersson)
(Machine issues: GENIVI IT mgr Nicholas Contino)

Standard Docker setup

Large machine, Yocto builds, SSD cache
go-agent2GENIVI Data Center
see above
Standard Docker setup

same specs as above


Standard Docker setup

Large machine, Yocto builds


Specialized agent for Vagrant/VirtualBox jobs


IT responsible:      Taha Mohammad
or via:                   Oscar Andreasson

Standard Docker setup

Large machine, Yocto builds
deltaStandard Docker setup
(intermittent use)

Yocto builds (32 threads, fast)
elmer(intermittent use)Yocto builds (slow)



(Go issues : Gunnar Andersson)
(Machine issues: Roland Krause)


Large machine, Yocto builds

NOTE:  This agent only runs between 12AM and 8AM, Pacific Time.
So, around 1700 CET it might interrupt a build if something is running at that time.
Let's see what effects this has to run start/stop an agent like that - please report findings.

Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz, 16GB RAM


Modifying and adding pipelines will require us to keep good communication.

  • Maintainers (see above) have superuser (Go Administrator) privileges.
  • All others that have editing permissions can also create and modify pipelines.  You are encouraged to experiment in the Experiments_and_Tests pipeline group.  If the pipeline starts working, it can then be moved to another group.
  • One way to keep up-to-date with changes is to track the continuous backup of the go-server configuration on GitHub (If you clone and pull this repository regularly you will be helping to keep a backup copy of the most relevant settings).   Every change to pipeline configuration is committed to git and the server pushes it to the backup remote on GitHub once per hour.
  • When in doubt, ask, for example on  If there is significant discussions we may consider a mailing list for coordination.

Creating Pipelines

Detailed information about each pipeline and some processes for creating new pipelines can be found in:

Agent Resources

Resources are a way to specify what an agent provides. 

Jobs that run as part of a pipeline stage can require particular resources and will then only be assigned to an agent carrying that resource.

Note that jobs that don't ask for any special resource can be assigned to any agent.

First, every agent should be tagged with a resource equal to the agent's host name.  This allows us to easily schedule certain administration jobs to specific agents.

Resource NamePurpose and explanation

Example of Pipeline
needing this resource

Example of Agent
having this resource


This signifies an agent machine that has been entrusted to keep a copy of licensed binary drivers for Renesas hardware, typically graphics drivers.  Without those binaries the build would fail but since they are only distributed after user license acceptance, there is no opportunity for an Agent to download them as part of the pipeline input material.

Special build steps are included to copy the binaries from a predefined location on an agent machine carrying this resource.

All GDP / Baseline builds for Renesas Hardware

Most Yocto_build enabled agents – please refer to the actual agent list on server (might need admin credentials)


This is used to indicate the agent is capable of running full Yocto builds of GDP size. 

This might go away if all agents are capable of doing that, or if those that are not (for example on-target test agents) are instead placed in a unique Environment (see below).

Full GDP / baseline builds using Yocto.

Most agents, unless they are known to be small (slow, or with little disk space).


This resource is specified on an agent dedicated to control a board farm for <architecture>, i.e. testing jobs can be assigned to run on it.

GDP & Baseline automated tests
arm (N/A)This is an agent machine that runs an ARM processor appropriate for native compilation/execution on ARM.

currently none used


Environments are similar but a more strict way to subdivide part of the Go work into sections.  An Agent can belong to maximum one environment and if it does, it will only run pipelines that explicitly belong to that environment.  That is typical for deployment where you want to direct certain pipelines strictly to a particular machine, and no other pipeline type should run on that machine.

No environments are used at this time.


Parameters are used to parameterize some pipeline templates.  Refer to documentation for explanation.


Pipeline templates are used heavily and encouraged.  This reduces repetition of the build steps for pipelines that only differ in Material (git repo or branch) or configuration (using Environment variables or Parameters). documentation

GitHub integration

Two plugins are used: Build GitHub pull requests and GoCD build status notifier.

These will build Pull Requests as soon as they are created and indicate directly in the Pull Request if the build was successful ((warning) limitations below).


Most of the information is in the README of the plugins (linked above) and won't be repeated.  Here follows information specific to our installation:


The plugin to build pull requests uses an OAuth token that it fetches from a file: $HOME/.github on the go-server.  (In our instance $HOME is /var/go)
The token is valid for the genivigo login name.

The other plugin that updates Github's status has its configuration under the Go-Server Plugin Admin interface (web interface).
Also there the same genivigo user OAuth token is used.

The user that is allowed to write the status of checks (genivigo) must be a collaborator of the project, and it seems it must have write access role.  (If you set this up on your own, remember to limit the OAUTH token access as much as possible, except for repository access)

(warning) Bugs: It used to be that ony PRs for master worked.  This is reportedly fixed now.

HOWTO for maintainers:

To build pull requests all you need is the first plugin.  To also update PR status you need to add the GitHub user genivigo as collaborator in your repository as above. 

Then set up a separate pipeline according to instructions in the Readme:  

  • As you can see in the README it's a little complicated to add the pipeline because you need to create a pipeline (which immediately forces you to add one Material but GitHub is not given as a choice there).  After saving, you go back and edit the pipeline to add new material of type "GitHub".  And then you finally remove the original material.  There are some hoops to jump through (make sure a subdir is defined for each material during the time you have multiple materials, and then you may want to remove the subdir usage if only one Material at the end) 
  • Once that is done, run the pipeline once manually (it will then build master branch, see below).  After that it should pick up any new pipelines.  If it doesn't work, use the Readme or ask that project.

Some deeper information about how pipeline design affects GitHub reporting is available in: TOOL-86 - Getting issue details... STATUS

  • No labels