Sorry about that - brown paper bag time...
swappiness was (coming from 2.6.8) still set to 0. As usual setting it back to
its default of 60
fixes the problem...
Would it be worthwhile to include the swappiness value in the OOM-killer's
report?
Wolfgang
Wolfgang Wander wrote:
Something still seems not right with the OOM handling.
The appended program forks 5 children, allocating 1GB each. Running
this on a 4GB RAM Intel32 machine that also has 11GB swap available
causes an OOM killer activation killing off one of the memory hungry
kids. (A 4GB AMD64 machine with 11GB
Something still seems not right with the OOM handling.
The appended program forks 5 children, allocating 1GB each. Running
this on a 4GB RAM Intel32 machine that also has 11GB swap available
causes an OOM killer activation killing off one of the memory hungry
kids. (A 4GB AMD64 machine with 11GB
Sorry about that - brown paper bag time...
swappiness was (coming from 2.6.8) still set to 0. As usual setting it back to
its default of 60
fixes the problem...
Would it be worthwhile to include the swappiness value in the OOM-killer's
report?
Wolfgang
Wolfgang Wander wrote:
4 matches
Mail list logo