|
|
Sponsored Content
Top Forums
Shell Programming and Scripting
executing a script for a certain amount of time
Post 28859 by Perderabo on Wednesday 25th of September 2002 01:15:58 PM
09-25-2002
I'm a little disappointed that the OP didn't bother to deny breaking our rules by posting a homework question.
But this turns out to be much harder than it first appears. I had a hard time doing this right and I must suspect that the OP's instructor underestimated the problem. So I am going to bend our rules a bit and offer a few hints on how to approach this.
If the subsidiary script is a script that never invokes another process, never forks and simply uses shell built-ins, this would be a very trivial problem. But if the subsidiary script invokes other processes, we really need to stop and restart all those processes in one fell swope. This implies working with process groups instead of processes. And ksh, the shell I used, has very limited support for process groups. In fact, I believe that I am relying on undocumented behavior of ksh. But I have tested this under HP-UX and it is working.
To get the master script to even employ process groups I had to do:
set -o monitor
early in the master script. Turning the monitor option on and off is really intended for interactive shells. I seemed to get away with it in a script. Most of the job control stuff did not work very well in non-interactive mode. I was hoping that I just just use the job control commands in a script, but there were too many problems with them. But with monitor mode set, the shell put background jobs in a seperate process group and it set the process group id to the process id returned in the $! parameter. I expected this behavior, but I can't find it documented on the ksh man page.
But now I had both the pid and process group id of the background process. The kill command can take a negative number as the second argument. If it is not -1, it is taken as a process group id and the signal is sent to all members of the process group.
The signal SIGSTOP will cause a process to suspend. And the signal SIGCONT will cause a stopped process to resume.
When the subsidiary process finishes, the process group will persist until all members of the process group have exited. At that point, a kill command target to the now defunct process group will fail. A script can check the return code fro the kill command to detect when this happens.
This should be enough clues to complete the homework assignment. Have fun... I did!
But this turns out to be much harder than it first appears. I had a hard time doing this right and I must suspect that the OP's instructor underestimated the problem. So I am going to bend our rules a bit and offer a few hints on how to approach this.
If the subsidiary script is a script that never invokes another process, never forks and simply uses shell built-ins, this would be a very trivial problem. But if the subsidiary script invokes other processes, we really need to stop and restart all those processes in one fell swope. This implies working with process groups instead of processes. And ksh, the shell I used, has very limited support for process groups. In fact, I believe that I am relying on undocumented behavior of ksh. But I have tested this under HP-UX and it is working.
To get the master script to even employ process groups I had to do:
set -o monitor
early in the master script. Turning the monitor option on and off is really intended for interactive shells. I seemed to get away with it in a script. Most of the job control stuff did not work very well in non-interactive mode. I was hoping that I just just use the job control commands in a script, but there were too many problems with them. But with monitor mode set, the shell put background jobs in a seperate process group and it set the process group id to the process id returned in the $! parameter. I expected this behavior, but I can't find it documented on the ksh man page.
But now I had both the pid and process group id of the background process. The kill command can take a negative number as the second argument. If it is not -1, it is taken as a process group id and the signal is sent to all members of the process group.
The signal SIGSTOP will cause a process to suspend. And the signal SIGCONT will cause a stopped process to resume.
When the subsidiary process finishes, the process group will persist until all members of the process group have exited. At that point, a kill command target to the now defunct process group will fail. A script can check the return code fro the kill command to detect when this happens.
This should be enough clues to complete the homework assignment. Have fun... I did!
|
|
10 More Discussions You Might Find Interesting
1. Shell Programming and Scripting
I do not have the luxery of a server that i can try on, so i had to post my qn here.
Say i have two jobs in the cron table, the jobs are scheduled 2 mins apart. Assuming the first jobs does a lot of operations and takes > 2 mins. Will the second job be executed? Will Unix actually have a queue... (3 Replies)
Discussion started by: new2ss
3 Replies
2. Shell Programming and Scripting
HI ,
I have a list1 which consists of data that i have to search and a list2 which has the files that need to be searched .So basically i am using list1 on list2 to see if list1 data is present if found replace it .I have written the code using foreach loop for each list .This is taking the... (1 Reply)
Discussion started by: madhul2002
1 Replies
3. Shell Programming and Scripting
Hi,
Unexpectedly, the function below doesn't seem to work
when called a second time. The output shows that when
the function is called the first time, it works as expected
but when it is called a second time, the loop that processes
the options passed to the function, with the while loop,... (2 Replies)
Discussion started by: ASGR
2 Replies
4. Programming
Hi All,
This Monday 15th March 2010, i have faced a weired issue with my Perl script execution, this script is scheduled to run at 1 minute past midnight on daily basis ( 00:01 EST ) generally for fetching previous business date , say if it is Monday it should give last Friday date, for Tuesday... (0 Replies)
Discussion started by: ravimishra
0 Replies
5. Shell Programming and Scripting
Hello all,
Hopefully someone can point me in the right direction...
I have a script written in bash which is pretty basic and just stop/starts various services based on particular conditions. What I am trying to build is a reporting type function which will send out an email with various stats... (3 Replies)
Discussion started by: systrex
3 Replies
6. Shell Programming and Scripting
hi,
i have written an shell script which reads an folder and write the filenames to another file:
here is the script:
#!/bin/bash
find /var/www/vhosts/remixland.de/web_users/stream001/jingle15 -type f -name "*.mp3" > /etc/shoutcast001/example.lst
killall -USR1 sc_trans_linux001... (1 Reply)
Discussion started by: maydo
1 Replies
7. UNIX for Advanced & Expert Users
I would like to kill a process after a certain amount of time. Can I please get some ideas on how to do this? (9 Replies)
Discussion started by: cokedude
9 Replies
8. Shell Programming and Scripting
Hi
Am having file.ksh as below
wc -l file1.txt
wc -l file2.txt
wc -l file3.txt
wc -l file4.txt
i want all the commands in this file to execute in same time
please help
Thanks in advance (1 Reply)
Discussion started by: ragu.selvaraj
1 Replies
9. Shell Programming and Scripting
Hi,
I have been trying to do a small fun project for myself.
I want to run a command for 45 seconds.
And to get the final output of this command, the script requires I push the "q" key on my keyboard and then the final output file becomes available.
I tried the following script. But it... (12 Replies)
Discussion started by: jacobs.smith
12 Replies
10. Shell Programming and Scripting
I have a script that checks if the script has been ran with sudo.
If the script is not ran as sudo, the current script is being executed with exec sudo bash.
You are asked for a password, you type in the password, success. Everything is perfect - the commands inside the script are ran as sudo.... (1 Reply)
Discussion started by: boqsc
1 Replies
LEARN ABOUT REDHAT
catch
catch(n) Tcl Built-In Commands catch(n) __________________________________________________________________________________________________________________________________________________ NAME
catch - Evaluate script and trap exceptional returns SYNOPSIS
catch script ?varName? _________________________________________________________________ DESCRIPTION
The catch command may be used to prevent errors from aborting command interpretation. Catch calls the Tcl interpreter recursively to exe- cute script, and always returns without raising an error, regardless of any errors that might occur while executing script. If script raises an error, catch will return a non-zero integer value corresponding to one of the exceptional return codes (see tcl.h for the definitions of code values). If the varName argument is given, then the variable it names is set to the error message from interpret- ing script. If script does not raise an error, catch will return 0 (TCL_OK) and set the variable to the value returned from script. Note that catch catches all exceptions, including those generated by break and continue as well as errors. The only errors that are not caught are syntax errors found when the script is compiled. This is because the catch command only catches errors during runtime. When the catch statement is compiled, the script is compiled as well and any syntax errors will generate a Tcl error. EXAMPLES
The catch command may be used in an if to branch based on the success of a script. if { [catch {open $someFile w} fid] } { puts stderr "Could not open $someFile for writing $fid" exit 1 } The catch command will not catch compiled syntax errors. The first time proc foo is called, the body will be compiled and a Tcl error will be generated. proc foo {} { catch {expr {1 +- }} } KEYWORDS
catch, error Tcl 8.0 catch(n)