I have recently upgraded my Amazon cloud t2.micro instance from Ubuntu 15.04 to 15.10. And soon I noticed a very strange high load on the machine - constantly above 2.0, and without doing much. The apparent culprit seemed kswapd0 process. I tried various simple steps to reduce the load, but nothing seemed to work, mainly because, as I already mentioned, no heavy activity was supposed to be happening.
After some Internet searching I came up with this bug report: kswapd0 100% CPU usage . Luckily, there is a suggested workaround from one of the contributors, although there is no permanent fix yet:
- Comment out line 2
(ATTR{[dmi/id]sys_vendor}=="Xen", GOTO="vm_hotadd_apply")in
/lib/udev/rules.d/40-vm-hotadd.rules/lib/udev/rules.d/40-vm-hotadd.rulesto
/etc/udev/rules.d/, so that subsequent upgrades to udev don't override your changes
This should take care of this particular issue. I applied the fix a few days ago and the load seems to be stable, in line with the expected usage levels.
It is worth noting, that Ubuntu 16.04 seems to be afflicted by the same problem. Just apply the solution outlined above and it should fix it.