Linux怎么正确捕捉内核崩溃信息

首先安装必选包：

apt-get -y install aptitude kdump-tools crash kexec-tools makedumpfile linux-image-`uname -r`-dbgaptitude full-upgrade # 避免运行的内核版本与调试的版本不一致导致无法调试

Kdump配置文件 /etc/default/kdump-tools 关键部分

USE_KDUMP=1KDUMP_SYSCTL="kernel.panic_on_oops=1"KDUMP_KERNEL=/boot/vmlinuz-3.16.0-4-amd64KDUMP_INITRD=/boot/initrd.img-3.16.0-4-amd64KDUMP_COREDIR="/data/crash"KDUMP_FAIL_CMD="reboot -f"DEBUG_KERNEL=/usr/lib/debug/vmlinux-3.16.0-4-amd64MAKEDUMP_ARGS="-c -d 31"KDUMP_CMDLINE="crashkernel=512M"

crashkernel大小配比，正常情况崩溃后1分钟左右会自动重启、配置不正确会导致重启卡住黑屏不动。

内存大小 crashkernel=0 - 12G 128M13 - 48G 256M49 - 128G 512M129 - 256G 1G *(896M, 768M o或 512M)

Grub配置文件/etc/default/grub关键部分

GRUB_CMDLINE_LINUX_DEFAULT="nmi_watchdog=1 crashkernel=512M"

Sysctl配置文件/etc/sysctl.conf关键部分

kernel.sysrq = 1kernel.watchdog = 1kernel.nmi_watchdog = 1kernel.panic_on_oops = 1kernel.softlockup_panic = 1kernel.watchdog_thresh = 10

重启让配置生效

update-grubreboot -fkdump-config loadkdump-config show

校验

# cat /proc/cmdlineBOOT_IMAGE=/boot/vmlinuz-3.16.0-4-amd64 root=UUID=a58ab901-00aa-4f8b-b3eb-d352fc72233 ro net.ifnames=0 thash_entries=1048576 rhash_entries=1048576 biosdevname=0 nohz=off enforcing=0 ipv6.disable_ipv6=1 nmi_watchdog=1 selinux=0 transparent_hugepage=never cgroup_enable=memory swapaccount=1 vga=771 crashkernel=512M# kdump-config testUSE_KDUMP:         1KDUMP_SYSCTL:      kernel.panic_on_oops=1KDUMP_COREDIR:     /data/crashcrashkernel addr:  0x2e000000kdump kernel addr:kdump kernel:  /boot/vmlinuz-3.16.0-4-amd64kdump initrd: /boot/initrd.img-3.16.0-4-amd64debug kernel: /usr/lib/debug/vmlinux-3.16.0-4-amd64kexec command to be used: /sbin/kexec -p --command-line="BOOT_IMAGE=/boot/vmlinuz-3.16.0-4-amd64 root=UUID=a58ab901-00aa-4f8b-b3eb-d352fc7f6acb ro net.ifnames=0 thash_entries=1048576 rhash_entries=1048576 biosdevname=0 nohz=off enforcing=0 ipv6.disable_ipv6=1 nmi_watchdog=1 selinux=0 transparent_hugepage=never cgroup_enable=memory swapaccount=1 vga=771 irqpoll maxcpus=1 nousb systemd.unit=kdump-tools.service crashkernel=512M" --initrd=/boot/initrd.img-3.16.0-4-amd64 /boot/vmlinuz-3.16.0-4-amd64

测试

echo c > /proc/sysrq-trigger

分析

crash /data/crash/201609010252/dump.201609012233 /usr/lib/debug/lib/modules/3.16.0-4-amd64/vmlinux

开始欢快的调试之旅

常用错误 ：

WARNING: kernel version inconsistency between vmlinux and dumpfile # 版本不一致，full-upgrade 一下。崩溃后系统不重启。 # 请检查 sysctl 、 crashkernel 及 nmi_watchdog 的设定！