On Sun, Mar 23, 2008 at 3:19 PM, Ali Saidi <[EMAIL PROTECTED]> wrote:
>  >>> scons: *** [build/SPARC_SE/tests/fast/long/00.gzip/sparc/linux/o3-
>  >>> timing/stdout] Error 3
>  >>
>  >> The pool node just hung up on the connection... I'm rerunning on
>  >> zizzer directly to see what I get there.
>  >
>  > These glitches seem to happen a lot.  What failures are most common
>  > tests blowing up or compile failures?  I've developed a better way to
>  > use a simulation pool for compiling using distcc that could easily be
>  > ported to OAR. It should at least fix the compile problems since
>  > distcc can recover from errors.  Anyway, I've got to get my stuff
>  > working on our simulation farm, so if anyone can give any pointers as
>  > to why this fails, I can probably fix them
>  The problem is oar closing the interactive shell before gzip finished.
>  I imagine it only effects gzip because it's one of the longest running
>  benchmarks we have. The release notes for a new version of oar say
>  that a bug was fixed with walltime. I upgraded oar and specified wall
>  time in a more standard manner. I guess we'll see if that fixes it.

Great, thanks.  FWIW I just ran it directly on zizzer and it passed
(after 6.5 hrs of CPU time) so it's definitely an OAR thing and not an
m5 thing.

Steve
_______________________________________________
m5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/m5-dev

Reply via email to