Memory problems on Redhat Enterprise 4
We have a Redhat Enterpise 4 server (Opteron dual core)
kernel 2.6.10 with two sata-II marvell controllers.
It is connected via 2 port aggregated network controllers to the network.
For some reason, whenever we perform massive copies (10GB or so) via ssh, the kernel starts running the oom service, resulting in the slow crash of the server (process after process)
Is there a way to stop the oom on 2.6.10 ?
Does anyone know the reason for this behaviour ?