soft and hard limit

soft限制了资源使用上限; soft可调整; hard限制了soft上限; 普通用户可使用ulimit -H调低hard limit.

限制的是一个进程可用资源, 而不是某个用户总和.

man setrlimit

RLIMIT_MEMLOCK

This is the maximum number of bytes of memory that may be locked into RAM.

RLIMIT_NOFILE

This specifies a value one greater than the maximum file descriptor number that can be opened by this process.

https://www.gnu.org/software/bash/manual/html_node/Bash-Builtins.html

ulimit provides control over the resources available to processes started by the shell, on systems that allow such control.

临时修改

ulimit命令默认查询/修改soft, 除非明确指定-H. 普通用户不能使用ulimit调大hard, root可以.

$ ulimit -S -n
1024
$ ulimit -H -n
4096
$ ulimit -H -n 2048
$ ulimit -H -n
2048
$ ulimit -S -n
1024
$ ulimit -n 512
$ ulimit -S -n
512
$ ulimit -H -n 4096
-bash: ulimit: open files: cannot modify limit: Operation not permitted
$ ulimit -H -n 3000
-bash: ulimit: open files: cannot modify limit: Operation not permitted

永久修改某个用户的limit

man limits.conf

vi /etc/security/limits.conf:

witness soft nproc 256
witness hard nproc 512
witness soft nofile 256
witness hard nofile 512

重新登录witness用户:

[witness@vm102 ~]$ ulimit -H -n
512
[witness@vm102 ~]$ ulimit -S -n
256
[witness@vm102 ~]$ ulimit -S -u
256
[witness@vm102 ~]$ ulimit -H -u
512

另一种方式: 把ulimit指令放在.bash_profile/.bashrc中[1][2].

proc文件系统可以看到进程当前的limits

$ cat /proc/23/limits
Limit                     Soft Limit           Hard Limit           Units
Max cpu time              unlimited            unlimited            seconds
Max file size             unlimited            unlimited            bytes
Max data size             unlimited            unlimited            bytes
Max stack size            8388608              unlimited            bytes
Max core file size        0                    unlimited            bytes
Max resident set          unlimited            unlimited            bytes
Max processes             1598                 1598                 processes
Max open files            1024                 4096                 files
Max locked memory         65536                65536                bytes
Max address space         unlimited            unlimited            bytes
Max file locks            unlimited            unlimited            locks
Max pending signals       1598                 1598                 signals
Max msgqueue size         819200               819200               bytes
Max nice priority         0                    0
Max realtime priority     0                    0
Max realtime timeout      unlimited            unlimited            us

nofile/ulimit -n

碰到这个错误: too many open files, 就是这个值不够, 或程序资源没释放. 如果需要可以调大一点.

给别人留一点: hard limit应该小于/proc/sys/fs/file-max[2][3]:

It is not recommend to set the "hard" limit for nofile for the oracle user equal to /proc/sys/fs/file-max. If you do that and the user uses up all the file handles, then the entire system will run out of file handles. This may prevent users logging in as the system cannot open any PAM modules that are required for the login process. That is why the hard limit should be set to 63536 and not 65536.

/var/log/messages: VFS: file-max limit 131072 reached

由于这个是控制的单个进程的资源使用, 一些场景如多进程+多文件, 即每个进程都要打开那么多文件, 就容易超过file-max, 比如进程数200, 文件数500, 200x500=100000 > 65535. 这时候需要调整file-max.

当前系统使用的fd总数: cat /proc/sys/fs/file-nr

进程使用fd数排序, 结果第二列是pid: lsof -n | awk '{print $2}' | sort -nr | uniq -c | sort -nr | more

core/ulimit -c

如果ulimit -c是unlimited, 但是无法生成或找不到core文件, sysctl -a | grep core_pattern.

参考

[1] https://access.redhat.com/solutions/61334

[2] https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/5/html/tuning_and_optimizing_red_hat_enterprise_linux_for_oracle_9i_and_10g_databases/chap-oracle_9i_and_10g_tuning_guide-setting_shell_limits_for_the_oracle_user

[3] https://serverfault.com/questions/235059/vfs-file-max-limit-1231582-reached, "I believe this to be an NFS server bug."