On Thu, Sep 22, 2011 at 02:38:47PM -0700, David Rientjes wrote: > The problem is this: > > Sep 20 19:39:19 host2 kernel: [1933000.196980] [ pid ] uid tgid total_vm rss cpu oom_adj oom_score_adj name > ... > Sep 20 19:39:19 host2 kernel: [1933000.197559] [13918] 507 13918 17992270 7758558 4 -17 -1000 root.exe > > root.exe is has about 29.5GB of the 32GB available memory in RAM, and it's > set to have a /proc/13918/oom_score_adj of -1000 meaning it's not eligible > for oom killing. So the kernel panics rather than kill the task. > > There's not much the kernel can be expected to do in such a configuration, > you've simply exhausted all RAM and swap. You can set > /proc/pid/oom_score_adj to not be -1000 so that it is at least eligible to > be killed in these circumstances rather than panic the machine, but the VM > will continue to oom under this configuration. It's surprising that the same workload in 32-bit works. Manoj, is root.exe recompiled for 64-bit ? I'm wondering if it's just that the expansion of a lot of unsigned longs are causing increased memory use vs the original 32bit use-case. Dave -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/ Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>