[Nfd-dev] Jenkins slave resiliency

[Nicholas Gordon]

I don't see any problem with this kind of approach--it's a pretty
standard availability tactic. I don't know if this adds much overhead
to managing the build system, that would be a question for Eric.

-- 
-Nick

On Sun, 01 Apr 2018 02:04:45 +0000
Junxiao Shi <shijunxiao at email.arizona.edu> wrote:

> Dear folks
> 
> Right now, all Jenkins slaves hosted at CSU are down. CSU hosts all OSX
> 10.12 and 10.13 slaves. Therefore, none of the build jobs can complete.
> This is not the first time that all slaves of a certain OS are failing at
> the same time, and most likely this would not be the last.
> 
> How about deploying the nodes in a more resilient way?
> *For each OS version, there should be slaves in at least two different
> sites*.
> This ensures the build jobs can proceed when there is at most one failure.
> 
> Yours, Junxiao