Need help with cron job syntax. Thank you.

urglik · 09-07-2014, 06:14 PM

hello everyone,

i've been having some fun setting up a vps to host some sites for my friends.

in an effort to back up their sites i'm using cron jobs and grive to tar the site folder and sync it to google drive.

the problem that has come up is that syncing looks for new or different file names and the cron job that tars the site folder saves it with same name each time by overwriting the existing file.

the syntax i use is

tar -cvpzf /site_backups/purpleearth.tar.gz /home/purpleearth/public_html

is there a way to have the tar file saved with a postfix like 01, 02, etc. depending on the existing files in the directory?

thank you for sharing your experience.

suicidaleggroll · 09-07-2014, 06:28 PM

The easiest and most descriptive option is to postfix the date instead of a counter. That way not only do you know "3 came after 2", but you know when each backup was actually saved, and you never have to worry about duplicates.

To do that, just change your tar command to:

Code:

tar -cvpzf /site_backups/purpleearth_$(date -u +%Y%m%d_%H%M%S).tar.gz /home/purpleearth/public_html

Of course you can change the format code to whatever you wish, use "man date" for the available options. The "-u" flag uses UTC for the timestamping rather than your current local time.

unSpawn · 09-07-2014, 06:49 PM

Two things: don't use "-v" with tar (unless you need to capture and log output) and try to use a backup scheme that provides retention.

Code:

#!/bin/bash --
# Set debug mode when testing:
set -vxe
# Set default behaviour:
LANG=C; LC_ALL=C; export LANG LC_ALL
# Be nice:
which nice >/dev/null 2>&1 && nice -n +20 $$ >/dev/null 2>&1
which ionice >/dev/null 2>&1 && ionice -c3 -n3 -p $$ >/dev/null 2>&1
# This provides retention for a week and only backups newer files during the week:
NOW=$(/bin/date +%a); THEN=$(/bin/date +"%Y-%m-%d" --date="1 day ago")

case "$NOW" in
Sun) # Create full backup:
     tar -cpzf "/site_backups/purpleearth-${NOW}.tar.gz" /home/purpleearth/public_html
      ;;
*)   # Create backup relative to yesterday:
     tar --newer "$THEN" -cpzf "/site_backups/purpleearth-${NOW}.tar.gz" /home/purpleearth/public_html
     ;;
esac
# Now rsync stuff and exit
# rsync --whatever-args source dest;
exit 0

urglik · 09-07-2014, 08:08 PM

urglik · 09-07-2014, 09:48 PM

wow,

thanks for the timely replies.

i tried suicidaleggrol's suggestion straight-away and it's a good quick fix to where i was at.

thank you. it brought my backup plan together.

i'm really digging grive.

google drive is stable, free, expansive.

for a beginner there were a lot of potential options at first glance.

unspawn, always being the one to take it too the next level,

if i can get my head around it, your suggestion intrigues me.

please to ask a few questions...

- i'm not following the rsync command. i had thought rsync was another option besides grive.
is this a different approach to backing up to google drive you are suggesting?

- is this a cron job?

- why did you include the first line?

trying to run before i learn how to walk...

thank you very much for your assistance.

unSpawn · 09-08-2014, 12:27 PM

Quote:

Originally Posted by urglik

i'm not following the rsync command. i had thought rsync was another option besides grive.
is this a different approach to backing up to google drive you are suggesting?

No, by all means, if you use grive, keep using that.

Quote:

Originally Posted by urglik

is this a cron job?

If you dump it in /etc/cron.daily/ it could be.

Quote:

Originally Posted by urglik

why did you include the first line?

Since it is an executable shell script it should start with a shebang / hashbang (see http://en.wikipedia.org/wiki/Shebang_(Unix) or for gory details see for example http://www.in-ulm.de/~mascheck/various/shebang/) and exit politely (zero).