Delete a file older than 30 minutes old

[beaster]

I have a file called alarm_sms.txt I need a command that will delete the file only after it is older than 30 minutes by system time.

Can anyone help?

//beaster

 

[KenCunningham]

Perhaps you could write a script including a sleep 1800 (ie 30 x 60 seconds) command to 'wake up' and delete the file 30 minutes after it is created. How is the file created and could this sleep be included within the creation script so that essentially it self-destructs after 30 mins? Something like:

*********Script to create the text file**********

sleep 1800

if [ -f alarm_sms.txt ]
then
rm alarm_sms.txt
fi

*********Loop back to top to recreate*************

Does this help?

 

[beaster]

Yes it does, but what do you think about what I have below?

find /home/beaster/alarm_program -name "alarm_sms.txt" -atime +30 -exec rm {} \;

It seems to be okay......tell me what you think.

//beaster

 

[beaster]

Also, when I write to this file during the 30 minutes, can I preserve the file date? I want to avoid writing to the file, it updating the time it was written to, because I would never delete it if it kept updating the time stamp.

Basically, I want to write to the file for 30 minutes and delete it after that.

//beaster

 

[KenCunningham]

Beaster, unfortunately, the -atime option in find uses days as it's criterion, not minutes. As far as I know, there isn't an option in find to specify minutes elapsed. HTH.

 

[KenCunningham]

If my perception of what you're trying to do here is correct, you're anxious that the file doesn't grow too large? If that's the case, you could set up a cron script to tun every half hour to trim the file to the last few entries. For example:

tail -30 alarm-sms.txt > temp
mv temp alarm-sms.txt

would take the last 30 lines of alarm-sms.txt and write it to temp, then use this to overwrite the larger file.

Apologies if this isn't what you have in mind.

 

[ayjaycee]

Sound's like Ken's idea is the best one, with the possible addition of a check to see if the process that writes to the file is still running when the sleep times-out & do a "kill" on it if it is.

http://www.tanda.f2s.com

One by one, the penguins steal my sanity.

 

[beaster]

Actually, I have been vague and I apologize.

I run commands to a node and pull all the alarms out and redirect to a file. > alarm_sms.txt

I then send to a paging system.

If I have already sent this alarm, I do not want to see it again for 30 minutes (Annoying at 2:30 in the am)

So pull alarms, redirect them to alarm_sms.txt, cat alarm_sms.txt > sms.txt and ftp it over to another node to be sent out.

Next time the script runs 10 minutes later, if it see's that there are alarms in alarm_sms.txt, (using fgrep) it won't put it in the sms.txt file before it sends it.

Then after 30 minutes, it starts all over.

Sorry for being stupid and not explaining what I am actually doing the first few times.

 

[KenCunningham]

Still a little confused, but would it be possible to compare the number of lines in the file each time it is examined, and if it has changed, to send only those added since the previous examination. Sorry - thinking aloud as I have to be elsewhere now! Hope you get this sorted - I'll check in the morning.

 

[dbmsguy]

here is my thought... (i did this on an HU/UX)

1. get the num of min after the hour file was created

ll xxx |awk '{print $8}'| cut -f2 -d:

2. get the current min after the hour

date |cut -f2 -d:

3. subtract. if the result is negative then add 90 (60 min + your 30 min).

if the answer is 30 or more, then remove the file otherwise sleep for 60 seconds. do all of this in an endless loop. Paul

http://www.dbmsguy.com

 

[dbmsguy]

oops.... sorry,

point 1 gets the minute the file was created, not the hour. that would be just dumb. Paul

http://www.dbmsguy.com