After further investigation on IRC the following points were raised:
1. Non-vcpu threads in QEMU weren't being isolated. Libvirt can do this
using the <cputune> domain XML element. The guest can create a high
load if some QEMU threads are unconstrained.
2. The wait% CPU stat was causing confusion. It's the idle time during
which synchronous I/O is pending. High wait% does not mean that the
system is under high CPU load. detect-zeroes=on can take a
synchronous I/O path even when aio=native is used, and this results
in wait% instead of idle%.
I'm closing the bug.
** Changed in: qemu
Status: Confirmed => Invalid
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1687653
Title:
QEMU-KVM / detect_zeroes causes KVM to start unlimited number of
threads on Guest-Sided High-IO with big Blocksize
To manage notifications about this bug go to:
https://bugs.launchpad.net/qemu/+bug/1687653/+subscriptions
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
