Re: [Springframework-developer] FAILED BUILD: sparc-solaris1 (last 30 lines included)

[Darren D. <da...@da...>]

On Fri, February 4, 2005 11:46, Rob Harrop said:
> I tested this particular test on Windows XP and it runs in 2400 ms on m=
y
> machine. Switching back to CGLIB 2.0.2 makes the test run in 2380 ms -
> so I guess there is no real difference. Does anyone have access to a
> Solaris workstation/server where they can execute this test manually?
> The interesting thing is that the test that is failing is inherited fro=
m
> CglibProxyTests and it seems to have passed on there assuming the tests
> are executed in order.

Rob,

it's likely just to be a problem with the compile farm machines at SF bei=
ng
hugely overloaded as they tend to be quite often.  Disk and CPU resources=
 are
stretched beyond the limit on a lot of those machines.  They can sometime=
s
take up to 30 minutes to complete a clean build/test run - a task that ta=
kes
about 2.5 minutes on my Linux box.

Perhaps we can make the max allowed time configurable somewhere so that t=
he
scripts which run the CF builds can just write a suitable value into a pr=
ops
file or supply it as a JVM arg when starting the build?  Unless we can me=
asure
CPU time of the thread itself instead of diffing the system time elapsed.

Cheers,


--=20
Darren Davison
Public Key: #DD356B0D