The oracle database running on the current network suddenly reports an error (alert Log) One day: too supports open files in system. You need to expand the number of allowed handles in the operating system and read a lot of information, the modification points are mainly concentrated in the following files:
1./proc/sys/fs/file-max
2./etc/sysctl. conf
3./etc/security/limits. conf
There are also methods to be modified through commands, such as ulimit-n 65535
In my actual experience, I modified/etc/security/limits. conf,
Oracle soft nofile 2048 oracle hard nofile 32768 oracle soft nproc 2048 oracle hard nproc 32768
The following describes the meanings of these values: 1. the first field is a restriction on the specified user. If it is changed to an asterisk, it indicates that it applies to any user. in the second field, soft is a soft limitation, and hard limitation is hard. It is too difficult to say so. What does it actually mean? For oracle users, you can use commands to modify your limits. However, the limit value cannot exceed the value corresponding to hard. If you do not change the value, the default limit is the soft value. (You can modify it through ulimit-n.) 3. nofile limits the number of handles that a single process can use. You can use lsof to view the number of handles used by a specific process, lsof checks that the number of handles used by the process must be doubled to the value corresponding to nofile. What is the relationship between them? 4. nproc limits the number of processes that can be started by the user. In actual tests, the number of processes viewed through the ps command is smaller than the configured value.
All in all, through the above configuration, you can change the number of processes that can use handles and the number of processes that can be started to limit.