river-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jonathan Costers <jonathan.cost...@googlemail.com>
Subject Re: Build failed in Hudson: River-trunk-QA #41
Date Thu, 04 Nov 2010 13:31:20 GMT
2010/11/3 Sim IJskes - QCG <sim@qcg.nl>

> On 11/03/2010 09:12 PM, Jonathan Costers wrote:
>> Oops. Will it really matter? River should survive in a chaotic network
>>> environment shouldnt it?
>>>  I've never tried to run many QA builds on the same network myself,
>> indeed it
>> would be interesting to see what happens.
>> One consideration would be the amount of consumed Hudson resources. We
>> would
>> be claiming one out of 2 ubuntu and one out of 2 solaris builder instances
>> for a long time ... Claiming either one or the other at the same time
>> would
>> spread that consumption more evenly.
>> Also, some of these tests probably assume a certain environment to be
>> present on the network.
>> For instance, a test may assume a number of Lookup services to be
>> discovered
>> on the network (because it knows how many it started itself)?
> Valid concerns. The solaris job does not run automatically (yet). Do you
> see possibilities to mutually exclude jobs with the hudson locking system?
Not 100% sure with the current locks that are in place.

There are locks for long running jobs on each machine, maybe we can assign
both the one for long running jobs on Ubuntu as well as the one for long
running jobs on hudson.zones.apache.org (Solaris) to both the QA builds.

We may benefit from a custom "River-QA" lock, and assign it to both QA

> Gr. Sim

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message