Scaling Jenkins Horizontally webinar questions and answers.

I presented a webinar yesterday titled “Scaling Jenkins Horizontally.” The webinar talked about using Jenkins Enterprise by CloudBees and Jenkins Operations Center by CloudBees products to build a resilient and scalable Jenkins infrastructure. This is a follow-up blog from the Q&A discussion during the webinar.

Resources:

These resources were mentioned by me during the webinar:

  1. User guide for Jenkins Enterprise by CloudBees (link)

  2. Reference architecture (link)

Question and Answers

Q: Is Jenkins Operations Center by CloudBees a standalone tool or requires Jenkins Enterprise by CloudBees?
A: Jenkins Operations Center by CloudBees = Jenkins + a set of plugins (plugins’) installed on Jenkins.
Downstream Jenkins (J’) = Jenkins + another set of plugins (plugins”) installed on Jenkins

Technically: Jenkins Operations Center by CloudBees manages J’. License wise J’ is considered Jenkins Enterprise by CloudBees.
Thus, Jenkins Operations Center by CloudBees is a standalone tool to manage Jenkins Enterprise by CloudBees.

Q: What happens to running jobs when Jenkins fails?
A: The jobs are lost and have to be restarted again. The restarted aborted builds plugin in Jenkins Enterprise by CloudBees, shows the admin a list of jobs that were running to quickly restart them.

Q: What is the licensing model for Jenkins Operations Center by CloudBees?
A: Jenkins Operations Center by CloudBees has its own license, plus a license for all masters and executors it manages. It takes over license management for individual masters from this global pool.

Q: We have one user scenario of several Jenkins masters and each master needs to have the same Jenkins jobs. Can we maintain all of those jobs from one location?
A: There is no special provision for that in Jenkins Operations Center by CloudBees at this time, though we would consider this kind of feature for the future.
(continuation of the previous question)

Q. We would like to have multiple jobs on each Jenkins server, and, for example, job1 on each Jenkins server has the same configuration, job2 on each Jenkins server has the same configuration, etc.
A. We can get there with the use of templates feature from Jenkins Enterprise by CloudBees, but are not there yet.

Q: Our security restricts how we connect to slaves. We cannot open a shell, but must issue a command like “java node/slave.jar.” Would that break shared agent ability?
A: You ought to be able to share agents using any launcher method available on both Jenkins Operations Center by CloudBees and attached masters, except that JNLP slaves are not yet supported.

Q: Is “haproxy” the only supported load balancing technology?
A: No, it is just a suggested integration. It is not exactly “load balancing” in this context, but rather hot failover.You can use JE/JOC HA without it, for example, by reassigning a DNS name or IP address to the currently active server.

Comments

  • “Shared slave is very handy” - yes it is!

Thanks for attending the webinar!

- Harpreet Singh
@singh_harpreet

 

 

Add new comment