On Mar 27, 2012, at 2:26 AM, Alan Bateman wrote:

> On 27/03/2012 05:21, David Holmes wrote:
>> 
>> It's a little disconcerting that 10 seconds is not enough time, but we have 
>> so many timing issues these days it is not that surprising. :(
> The machines in question are relatively beefy but there is hardware 
> visualization in the picture that may be an issue.
> 

Yup. Or Zones.  The hardware itself is dedicated to the testing, so it should 
have plenty of horsepower.

> Chris - another thing that is worth checking is whether any previous tests 
> are leaving looping threads. These tests are running sequentially in the same 
> VM and we have had cases in the past where a test creates a daemon thread 
> that heads off into the weeds, leading to slow down in subsequent tests. I 
> tracked down a few of these last year and don't think we have any now but 
> worth re-checking.

Yup. Could be a major impact being run as a jtreg -agentvm batch.

> 
> In any case, the test change looks fine to me too.

Yup. Me too.

-kto

> 
> -Alan.

Reply via email to