Delayed command?

[wubrgamer]

I'd like to (on my dreamhost shared debian server):

run a delayed command, I would like make a file available for download, but only for a short amount of time (say 48 hours) and then after that, remove the file. without having to manually removing the file? (as I have been doing through an FTP client.)

is there a way to do this?

I could put into my crontab "rm /home/USERNAME/SITE/file.ext" for the date and time that I would like to remove it from my server.

But how do I do this?

What would you do?

thanks for your help!

[wubrgamer]

basically, what I'm trying to do is auto-prune these files.

based on time.

[ocicat]

Originally Posted by wubrgamer

...I would like make a file available for download, but only for a short amount of time (say 48 hours) and then after that, remove the file. without having to manually removing the file? (as I have been doing through an FTP client.)

Set up a cron job which compares the timestamp when the file was uploaded to the current time. HINT: look at the +%s switch of date(1) which will return the number of seconds since the epoch.

• Substracting the upload timestamp from the current timestamp will yield the number of seconds lapsed since uploading.
• Convert 48 hours into seconds & if the lapsed time is greater than this threshold, delete the file.
• Iterate over all files you want to make available for only a prescribed amount of time.

This is one way you might do it within a shell script. Perl or Python will have modules which can look at elapsed time as well.

[J65nko]

I haven't used Linux for a long time, but doesn't crontab -e allow you to add a task to your personal cron table?.

For an automatic pruning after 48 hours you could use the find command. The following shows a listing of all files in my home dir older then file 48 hours

$ find /home/j65nko/ -ctime 2  2 -exec ls {} \;

Change the ls to rm -f and you have your auto-prune command

[jggimi]

The cron "system" usually includes an at(1) command.

Example:

$ echo sh bin/script | at 1am tomorrow
commands will be executed using /bin/ksh
job 1240290000.c at Tue Apr 21 01:00:00 2009
$

edit:

You can also use "now +" -- specific to your needs:

$ echo rm /path/to/file | at now + 48 hours

[wubrgamer]

thank you!

"at" is the PERFECT command for what I'm looking to do!

could you please help me find a primer or another place to start using "at"

[Albright]

at(1)?

[jggimi]

The man page is what I used; granted, it required a little experimentation before I figured it out. And, if I don't use it for six month or a year, I forget everything and have to start over, and learn it again.

At has two types of time specifications.

If you use "-t <timespec>", the "timespec" is a single format, starting with an optional century, then optional year, then month, then day, then hour, then minute, then optional seconds:

e.g: 200905012315.30 = this coming May 1, at 23:15:30. 05012316 = the same evening, 30 seconds later, at 23:16:00

The other formats that do not use -t are much more loosely defined.

[wubrgamer]

Yep, *at* is pretty awesome...

thanks!

[phoenix]

The find command via cron is much more automated (set it once and forget about it). But at is workable.

I prefer to use at for one-off type jobs, since the at-job is removed once it has executed. Whereas cron is better for continuous-type jobs (like the one you are asking about).