> On Feb 10, 2017, at 4:12 PM, Gavin McDonald <ga...@16degrees.com.au> wrote:
>
>
> I’m at a loss, I’ve given you positive responses to every question you posed,
> yet you cut those out of your replies and find something else to complain
> about. I suspect you will never be happy. Even with 27 nodes to play with and
> more on the way. I suspect you will never be happy, so be it, I’m done
> spending my
> time trying.
>
I'm not complaining. I'm trying to understand how the "social priority"
actually works in practice with the toolset put in place. If Hadoop (or HBase
or any number of other large projects currently sharing space on the 'Hadoop'
nodes) opens the firehose and starts running jobs on the general Ubuntu nodes,
build times for the rest of the ASF will absolutely tank when these large
projects get busy (especially around release time). People are very likely to
be extremely unhappy when they see one of our 8-10 hour monsters making their
one hour jobs take several hours (and maybe even mysteriously fail!). These
jobs are rarely kicked off manually, so we're very dependent upon how Jenkins
is enforcing queuing. If we as a community don't care that other projects are
getting crushed under the weight of an elephant ;), then so be it.
