背景
Linux KVM虚拟机由qemu加载和引导，虚拟机的运行内存也都映射在qemu进程空间里面，导致qemu的rss内存一般都很大，比如这两个qemu虚拟机，看起来差不多分别是16G和8G内存规格的，都使用了2G左右的物理内存。如果qemu进程由于任何bug发生core dump, 可能会导致产生一个非常巨大的core文件。
top - 12:46:02 up 6 days, 13:17, 5 users, load average: 1.14, 1.22, 1.72
Tasks: 532 total, 1 running, 279 sleeping, 0 stopped, 0 zombie
%Cpu(s): 0.1 us, 0.7 sy, 0.0 ni, 99.2 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st
KiB Mem : 19667006+total, 98877040 free, 5496464 used, 92296560 buff/cache
KiB Swap: 0 total, 0 free, 0 used. 18959481+avail Mem
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
3357981 root 20 0 17.087g 1.742g 22244 S 1.3 0.9 106:12.59 qemu-system-x86
4157696 root 20 0 9770.2m 1.474g 22676 S 2.0 0.8 6:38.13 qemu-system-x86
解决办法
编辑/etc/libvirt/qemu.conf, 下面的配置将会将qemu进程的core文件限制到2G, 并且不包括guest内存。
# Size is a positive integer specifying bytes or the
# string "unlimited"
#
#max_core = "unlimited"
max_core = 2147483648
# Determine if guest RAM is included in QEMU core dumps. By
# default guest RAM will be excluded if a new enough QEMU is
# This setting will be ignored if the guest XML has set the
# dumpcore attribute on the <memory> element.
#
#dump_guest_core = 1
dump_guest_core = 0
