Re: [hudson] Slave on EC2?

2010-12-16 Thread Niklas Gustavsson
On Thu, Dec 16, 2010 at 8:13 AM, Jukka Zitting jukka.zitt...@gmail.com wrote:
 Thus the need for my earlier
 idea (for which I also lobbied initial budget approval) of setting up
 an extra on-demand slave on EC2 for handling load peaks has mostly
 gone away.

The recent outage of Minerva has showed that we get long queues when a
slave is gone. Therefore, having a EC2 slave to activate in situations
like that would be good.

/niklas


Re: [hudson] Slave on EC2?

2010-12-15 Thread Matthias Wessendorf
What about CloudBees? (Hudson as a service)

Perhaps they grant OS orgs a discount? :)
(since the CEO is the formal JBoss CTO)

-Matthias

On Thu, Dec 16, 2010 at 8:13 AM, Jukka Zitting jukka.zitt...@gmail.com wrote:
 Hi,

 Despite various hickups along the way, the new Hudson master and the
 extra solaris slave we got from the old master seem to have gone a
 long way to addressing the chronic capacity problems we had earlier
 ([1] looks a lot better than it used to). Thus the need for my earlier
 idea (for which I also lobbied initial budget approval) of setting up
 an extra on-demand slave on EC2 for handling load peaks has mostly
 gone away.

 So the question is, should I still look at setting up an EC2 slave?
 Beyond load balancing, the other reason for running builds in the
 cloud is the availability of a wider variety of build environments.
 Basically we could provision an on-demand build slave for any
 environment for which an AMI exists or can be created. I don't have a
 pressing need for that from the projects I work with, but perhaps
 others are interested?

 [1] https://hudson.apache.org/hudson/load-statistics?type=hour

 BR,

 Jukka Zitting




-- 
Matthias Wessendorf

blog: http://matthiaswessendorf.wordpress.com/
sessions: http://www.slideshare.net/mwessendorf
twitter: http://twitter.com/mwessendorf