cron jobs are hard to make # prev single next top
cron jobs are a great tool, but it's very difficult to test them. a lot of the time, I create a cron job, and simply check back a day later to see if it did anything.
I've now developed a process (when making something work that's super annoying, you stick to your solution, even if it's hacky), which is:
-
make a bash script that does what I want the cron to do, e.g.,
collect.sh
-
cd
out of the folder and test the script works in a "sub-job" (brackets), i.e.,cd / (cd /usr/alifeee/cool-thing/; ./collect.sh >> cron.log)
-
put that exact command (with brackets) in the cron job
before I did this, I was making sure that the scripts I made worked when called from any directory (i.e., by making each call to a file use "${root_directory}/${file}"
). This was kind of annoying and I much prefer the "just cd
into the directory before running the script" method.
Here are all of my current cron jobs on my server currently:
0 10 * * * /home/alifeee/scripts/combine_treehouse.sh >> /home/alifeee/scripts/cron.log 2>&1
0 11 * * * /var/www/ringram/next.sh >> /var/www/ringram/cron.log 2>&1
0 23 * * * /var/www/ringram/update.sh >> /var/www/ringram/cron.log 2>&1
# clean up old calendar events
0 23 * * * /var/www/simple-calendar/hoover.py >> /var/www/simple-calendar/cron.log 2>&1
# new weather map every 15 mins
*/15 * * * * (cd /usr/alifeee/weather_landscape; /usr/alifeee/weather_landscape/.venv/bin/python /usr/alifeee/weather_landscape/make_image.py >> /usr/alifeee/weather_landscape/cron.log 2>&1)
# get latest sheffield events
0 3 * * * (cd /usr/alifeee/sheffield-events/; echo "" > cron.log)
0 4 * * * (cd /usr/alifeee/sheffield-events/; ./get_all.sh >> cron.log 2>&1)
some of the commands could really be simpler, but as I said: "if it works: I will leave it until it breaks"
they are related to these projects (match them up yourself):