Restart python script from crontab when it crash or killed

[syscharger]

Added the following in crontab (sudo crontab -e) to check for running of python script (testpython.py) every 5 minutes
If script not running, execute the script (testpython.py)..
But unfortunately, it does not work. Any idea why? Did i miss something. Please advise. Thanks

Code: Select all

*/5 * * * * pgrep -f testpython.py || sudo python /home/pi/testpython.py

[SonOfAMotherlessGoat]

Is the account this cron job runs under able to sudo without a password?

[syscharger]

Is the account this cron job runs under able to sudo without a password?

Yes. able to.
i tried similar code below, no problem.

*/5 * * * * sudo python /home/pi/testpython.py

[syscharger]

Is there alternative solution? If not using crontab.
Thanks

[DougieLawson]

Is there alternative solution? If not using crontab.
Thanks

Add a cron file in /etc/cron.d, if you have php5 installed there's an example file in there.

The simplest way to keep a python script running is

Code: Select all

#!/bin/bash

until (/home/pi/thingtokeepgettingrestartedwhenitdies.py);
do
sleep 30m # wait half an hour and give it another go.
done;

[rayjoh]

It might be that sudo denies cronjobs. Add this line in /etc/sudoers, substitute "rj" with your username.

Code: Select all

Defaults:rj   !requiretty

[alexeames]

don't you need to include the full path to python in your cron job?

cron can be exceedingly fussy

[QuietZone]

don't you need to include the full path to python in your cron job?

Nope.

cron can be exceedingly fussy

Yes, but this isn't one of them.

[jjex22]

Just a side note, that It also occurs to me that python is pretty good at exceptions, and some defensive programming combined with exploring recent crashes may help more in the long term, so that a resurrection script becomes more of a backup?

[mosespi]

As jjex22 mentioned you should try to make it more robust.. but a simple wrapper script using bash should ensure a restart if it does crash. Very simple code below, so simple its almost always going to work.. run it from rc.local if you wish, and keep the sleep in there just in case your script really hard crashes and refuses to start, it will keep it from restarting too fast and possibly using up most of your system resources.

And I missed reading Dougie's prior post.. this is a similar variant.

Regards,
-Moses

Code: Select all

#!/bin/sh

while [ 1 ]
do
        /home/pi/blablayourpythoncommand.py or whatever
        sleep 10
done

[jaizan]

Apologies for the dumb questions. I have PrivateEyePi running, but it drops out every few weeks. I'm trying to figure out how to make it restart if necessary.

1 Would something like the following work ?
2 Do I just put this in my CRON file ?
3 Is the example below just valid for something in the PI folder, so I need the correct path for restarter ?


#!/bin/sh

while [ 1 ]
do
/home/pi/restarter.py or whatever
sleep 10
done

[DougieLawson]

Use a systemd service file. /etc/systemd/system/privatepi.service

Code: Select all

[Unit]
Description=Private Pie server

[Service]
ExecStart=/usr/bin/python3 /home/pi/testpython.py
User=pi
Restart=always
RestartSec=360
StartLimitInterval=720
StartLimitBurst=10
StandardOutput=syslog
StandardError=syslog
SyslogIdentifier=PrivEye

[Install]
WantedBy=multi-user.target

If it absolutely must run as root remove the User=pi line (it probably doesn't need sudo most thing don't).

Enable that for the next boot with sudo systemctl enable privatepi
Start it now with sudo systemctl start privatepi

Test the restart with
sudo systemctl status privatepi
sudo killall testpython.py
sudo systemctl status privatepi