No matter how programs are not perfect, theoretical, any procedure has a day, officially operated procedures, especially server programs, once core dump, consequences, have experience in server development experience, must have experienced During the night, I was ambiguous experience that was awakened by the phone, and the hand was restarted, and I went to work on the next day. I was batch. Therefore, the automatic restart is important when the program is wrong. Here is a way to discuss Linux to implement an automatic restart program.
There are many ways to implement the program under Linux, here we introduce it to the book, automatically restart the script.
It is assumed that the program is required to be re-enabled, and we determine whether the program is normal by judging the number of processes.
PS-EF | GREP "$ 1" | grep -v "grep" | WC -L is the number of processes that get $ 1 (Test in this example), and the script determines the next step according to the number of processes. Check the number of processes in the system in the system every 1 second through a dead cycle.
The code is as follows: Script Check
#! / bin / sh # ------------------------------------------- ----------------------------------- # Function: CheckProcess # Features: Check if a process exists # parameters: $ 1 --- To check the name # Returns: If there is a return 0, return 1. # -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- -------------------------------------------------- CheckProcess () {# checks if the input parameter is valid if ["$ 1" = ""]; the return 1 Fi # $ process_num Get the number of the specified process name, to return 0, indicate normal, not 1 return 1, indicate There is a mistake, you need to restart process_num = `ps -ef | grep" $ 1 "| grep -v" grep "| wc -l`ness [$ process_num-eq 1]; the return 0 else return 1 fi} # Check Test instance Whether you already exist while [1]; do checkprocess "test" checkqq_ret = $? If [$ checkqq_ret -eq 1]; then # kills all TEST processes, can change the operation you need to do KILLALL -9 TEST EXEC. /TEST & Fi Sleep 1DONE
Script Start: Join the Limit Coredumpsize 102400, set the size of the Core File, once the program core dump, tracks can be found. Execute the Check script in the script in the background, you can save a lot of trouble,
#! / bin / cshlimit coreumpsize 102400./check &
