Linux查看程序被Killed的原因及分析
作者：Heerlen
文章总结：本文介绍了如何查看系统中被Killed的进程信息,并提供了设定kill优先度的方法,优先度通过oom_adj值来调整,范围为-16到+15,值越高优先级越高,文章还提到如何处理被拒绝访问的命令,并鼓励大家提供支持
Linux查看程序被Killed原因
1.查看信息，输出最近被killed的进程
查看被Killed简短信息列表 egrep -i -r 'killed process' /var/log
[root@ecs-bcb3 ~]# egrep -i -r 'killed process' /var/log /var/log/messages:Jul 12 15:00:24 ecs-bcb3 kernel: Killed process 26399 (python3), UID 1002, total-vm:27455044kB, anon-rss:24933916kB, file-rss:712kB, shmem-rss:0kB /var/log/messages:Jul 12 15:00:24 ecs-bcb3 kernel: Killed process 29412 (python3), UID 1002, total-vm:27455044kB, anon-rss:24934124kB, file-rss:780kB, shmem-rss:0kB /var/log/messages:Jul 12 15:18:39 ecs-bcb3 kernel: Killed process 25017 (chromedriver), UID 1002, total-vm:218888kB, anon-rss:2312kB, file-rss:184kB, shmem-rss:0kB /var/log/messages:Jul 12 15:18:39 ecs-bcb3 kernel: Killed process 30252 (chromedriver), UID 1002, total-vm:201472kB, anon-rss:2032kB, file-rss:104kB, shmem-rss:0kB
查看被Killed详细信息列表 dmesg | egrep -i -B100 'killed process'
[root@ecs-bcb3 ~]# dmesg | egrep -i -B100 'killed process' [19095552.852409] [31573] 1002 31573 131893 2522 166 0 0 chrome [19095552.852585] [31597] 1002 31597 144656 4247 168 0 200 chrome [19095552.852756] [31599] 1002 31599 148317 2766 183 0 0 chrome [19095552.869403] [26287] 1004 26287 1975489 186919 547 0 0 java [19095552.869548] [27054] 999 27054 1154 23 8 0 0 sh [19095552.869700] [27058] 999 27058 1110 16 8 0 0 df [19095552.869842] Out of memory: Kill process 7429 (python3) score 357 or sacrifice child [19095552.869991] Killed process 7429 (python3), UID 1002, total-vm:26420076kB, anon-rss:23508276kB, file-rss:920kB, shmem-rss:0kB ...
查看被Killed简短信息列表 journalctl -xb | egrep -i 'killed process'
12月 21 15:33:18 ecs-kms kernel: Killed process 2580 (chrome), UID 0, total-vm:862648kB, anon-rss:53044kB, file-rss:156kB, shmem-rss:2692kB 12月 21 15:33:18 ecs-kms kernel: Killed process 3000 (chrome), UID 0, total-vm:862136kB, anon-rss:49508kB, file-rss:16kB, shmem-rss:2508kB 12月 21 15:33:18 ecs-kms kernel: Killed process 3165 (chrome), UID 0, total-vm:859320kB, anon-rss:48372kB, file-rss:112kB, shmem-rss:2492kB ...
2.设定kill优先度
- 2.1 完全关掉oom
sysctl vm.overcommit_memory=2 echo “vm.overcommit_memory=2” >> /etc/sysctl.conf
- 2.2 出现 Out Of Memory 之后重启
sysctl vm.panic_on_oom=1 sysctl kernel.panic=X echo “vm.panic_on_oom=1” >> /etc/sysctl.conf echo “kernel.panic=X” >> /etc/sysctl.conf
- 2.3 优先考虑kill这个进程
sudo echo 10>/proc/[PID]/oom_adj
- 2.4 尽量不去kill这个进程
sudo echo -15>/proc/[PID]/oom_adj
- 2.5 不会kill这个进程
sudo echo -17>/proc/[PID]/oom_adj
备注：
- oom_adj的值在-16 到 +15之间，值越高被kill的优先度越高。
- 当该值为-17时，系统将不会杀死指定pid的进程，而-16~15则会使得进程的 /proc/[pid]/oom_adj 值呈指数形式递增，即它们被杀掉的可能性呈指数递增。
- 针对init（进程号为1）这个进程，无论该值设为多少都不会被杀
3.若上述命令被拒绝访问
命令修改成以下方式
bash -c "echo '10' | tee /proc/[PID]/oom_adj" bash -c "echo '-15' | tee /proc/[PID]/oom_adj" bash -c "echo '-17' | tee /proc/[PID]/oom_adj"
总结
以上为个人经验，希望能给大家一个参考，也希望大家多多支持脚本之家。