[users at bb.net] Best Practice use of workers

[Trevor Hook]

Hi

I'm just thinking about how I'd like to organise workers in the future.

The Background
--------------------------

I'm running Buildbot and 1 worker on a Raspberry Pi Zero W to create a couple of websites with Pelican.

I have the workers running in a virtualenv and that virtualenv also contains the Pelican package needed to build the websites, I did this for better portability, I just setup the worker with a script on any future machine and it should be good to go.

I'm starting to think of others things I could do with Buildbot and the obvious test I can think of is a check that changes to a Buildbot master config is valid.

The Question
---------------------

Does it make sense to update that worker virtualenv to also include Buildbot so the same worker can build websites or run Buildbot checkconfig or would you create a new worker which was only used for Buildbot type tasks?

Given the limited processing power of the RPiZeroW, I was planning on putting it in to the same worker and therefore limiting the system to only ever running 1 worker task at a time. 

However, it feels wrong mixing worker purposes like this. Is it possible to block a worker from running a job while a different worker from a different builder is running a job? That would allow me to keep the workers focused on 1 task while preventing the RPi from trying to do too much at once.

I'm just trying to get some idea of whether there is a great practice for this sort of configuration.

Thanks
Trevor
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.buildbot.net/pipermail/users/attachments/20171026/820855f6/attachment.html>