since some months now, we have regulary issues with memory managment, played around /proc/sys/vm/swappiness, but yet not found a fix. The kernel often isn't able to handle memory when it unused memory goes to 0...
if we use default (60), it swappes active memory and so IO goes crazy. We played around and also 0 (old behavior) seems not a solution. The server runs on Hetzner and is based on the Debian installation like described on the wiki here.
this munin graph should also show our issue:
https://munin.openttdcoop.org/openttdcoop.org/haydn.openttdcoop.org/memory.html
everytime, after reboot everything works fine again.
people from #openvz do not recommend to use the old .32 branch rahter use the stable branch for RHEL6, I am not sure, what this means exactly for you. But could it be this .32 kernel is a bit experimental?
Waht would you recommend for us? Do you need further infos?
Greets
Ammler
P.S.:
pve-manager/1.8/6070
Linux 2.6.32-4-pve #1 SMP Mon May 9 12:59:57 CEST 2011
aptitude does not recommend any update...
if we use default (60), it swappes active memory and so IO goes crazy. We played around and also 0 (old behavior) seems not a solution. The server runs on Hetzner and is based on the Debian installation like described on the wiki here.
this munin graph should also show our issue:
https://munin.openttdcoop.org/openttdcoop.org/haydn.openttdcoop.org/memory.html
everytime, after reboot everything works fine again.
people from #openvz do not recommend to use the old .32 branch rahter use the stable branch for RHEL6, I am not sure, what this means exactly for you. But could it be this .32 kernel is a bit experimental?
Waht would you recommend for us? Do you need further infos?
Greets
Ammler
P.S.:
pve-manager/1.8/6070
Linux 2.6.32-4-pve #1 SMP Mon May 9 12:59:57 CEST 2011
aptitude does not recommend any update...