A workaround that seems to work for me (and one that doesn't involve dropping caches periodically) is to downgrade udev to 219-7ubuntu6 (as suggested by Øystein above). On 15.10 wily, add vivid-updates to your apt sources to be downgrade to this version of udev.
I noticed this bug is more likely to occur on servers with less memory / more memory pressure. I was only able to reproduce it on our t2.micro and t2.small instances on EC2, and using larger EC2 instances was our first workaround. (In fact, after downgrading udev I was also downgrade to smaller EC2 instances again.) This bug also seems more likely to occur on servers with faster disk I/O, given I was only able to reproduce it on instances with gp2 (SSD) storage and not on standard (magnetic) storage on EC2. Our use case is a bunch of EC2 instances all using the Ubuntu 15.10 wily AMI, and currently running the latest 4.2.0-23-generic kernel. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/1518457 Title: kswapd0 100% CPU usage To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1518457/+subscriptions -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs