|
|
|
|
|||||||
| Forums | Portal | Register | Forum Rules | FAQ | Contribute | Members List | Arcade | Search | Today's Posts | Mark Forums Read |
| Shell Programming and Scripting Post questions about KSH, CSH, SH, BASH, PERL, PHP, SED, AWK and OTHER shell scripts here. |
More UNIX and Linux Forum Topics You Might Find Helpful
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| grepping for period . in file name ? ie grep . | bobk544 | UNIX for Dummies Questions & Answers | 5 | 11-30-2007 11:31 PM |
| output result (period -1) | happyv | Shell Programming and Scripting | 4 | 05-18-2007 03:11 AM |
| record wc -l over 24 hour period | csaunders | Shell Programming and Scripting | 5 | 12-09-2005 11:49 AM |
| Timeout period | Nadeem Mistry | UNIX for Dummies Questions & Answers | 4 | 08-15-2001 06:55 AM |
| Deleting a File with an Erroneous Period | PLF | UNIX for Dummies Questions & Answers | 3 | 08-08-2001 11:04 PM |
 |
|
|
Submit Tools | LinkBack | Thread Tools | Search this Thread | Display Modes |
|
#1
03-04-2006
|
|||
|
|||
|
Run job for a period of time
I have a job that runs for an unspecified amount of time. I want to run this as a cron job for a specified amount of time, say 2 hours. Once the time is up, the program should be killed in the middle of execution. How can I do this?
Thanks. 
|
| Forum Sponsor | ||
|
|
|
#2
03-04-2006
|
|||
|
|||
|
If your process is written in C and you have access to the code, you can use the alarm(2) function which will generate a SIGALARM signal after a specified number of seconds. You can capture the signal for a graceful exit or let it abort the program.
You can do the same with a script by sending a HUP signal to your script via at(1). At the beginning of your script, run the following command: echo "/bin/kill -HUP $$" | at now + 120 minutes With a trap, the HUP signal could be intercepted for a graceful termination or, as in the alarm(2) example, let it abort the script. Also, the at-job number could be saved used to remove it from the queue if your script completes early. A third option would be to create a pid-file from within your process (giving it a shell wrapper if necessary). Then you could cron another job to check for the existence of the pid file after a specified period of time and send a signal to it if the PID is still active. Here's kind of a not-too-useful example: Code:
#!/bin/ksh
trap 'echo terminated counts: $I : $J : $K ; exit 3' HUP
echo "kill -HUP $$" | at now + 1 minute
I=0
while [ $I -lt 1000 ]
do
J=0
while [ $J -lt 1000 ]
do
K=0
while [ $K -lt $J ]
do
(( K = K + 1 ))
done
(( J = J + 1 ))
done
(( I = I + 1 ))
done
echo final counts $I : $J : $K
exit 0
Last edited by hegemaro; 03-04-2006 at 07:02 PM.
|
|
#3
03-04-2006
|
|||
|
|||
|
Wow, that's just what I needed. I appreciate you going through with the effort with this.
I have some questions though: - why do I get the following message each time when I run the above script in succession: Job 2 will be executed using /bin/sh Job 3 will be executed using /bin/sh ... - in general, is there a preference of using SIGHUP over SIGINT in these situations?
|
|
#4
03-05-2006
|
|||
|
|||
|
this is one more way of implementing;
Code:
./yourscript.ksh & sleep 7200; kill $!
|
|
#6
03-06-2006
|
|||
|
|||
|
By default, the at command uses the Bourne shell. What you're seeing with "Job 2 will be executed using /bin/sh" message is a warning. You can call at with a -k option to tell it to use the Korn shell.
As far as SIGHUP vs SIGINT, it depends on the process you're trying to interrupt. Many processes handle SIGHUPs gracefully -- for example syslogd will reread its configuration file. Other process, if not coded specifically for processing SIGHUP will terminate. SIGINTs, on the other hand, are almost universally "process-terminating". Experiment. I'm glad I could be of help.
|
|||
| Google The UNIX and Linux Forums |
Linear Mode
