VM killing process on slave node (Scyld)

[James at armstrong.uk.net]

Hi,
        Recently I have upgraded from Scyld 27bz-7 to 27z-8 and found I can no
        longer run a job that requires swap on a slave node..

        Each slave has 512Mb of Ram and 1Gig swap space. This is shown by both
        bpsh 1 cat/proc/meminfo and bpsh 1 free. However if I run try and use
        more than 500Mb the process is killed and bpsh 1 dmesg gives VM:
        killing process x

        Any Ideas?

                                                        J.W.Armstrong
----------------------------------
E-Mail: James at Armstrong.uk.net
WWW:    http://www.armstrong.uk.net
----------------------------------