Folks,
I can’t seem to figure out this. I have a pre-compiled C program that I use to run some simulations. I would like to run that command whenever the BeagleBone (linux) boots.
I’ve done this a million times on ubuntu but for some odd reason I am having a hell of a time getting it to run on my board.
My normal approach is:
to go to the /etc/init.d directory and create a text file in there (e.g myScript.sh)
edit the script to something like
#!/bin/bash
/home/root/myProgram &
leave the script and add execution rights to it (e.g. chmod u=+rwx myScript.sh)
go to /etc/rc5.d and add a symbolic link to my script in there
ln -s …/init.d/myScript.sh S99myScript
This just doesn’t seem to work. Any ideas?
Also, how do I know what is the current runlevel since there is no inittab on the angstrom distribution.
I chmod’ed the script to 755 to make sure everyone could run it
the script as it’s first thing just creates a file via echo and them tries to launch two python programs.
#!/bin/bash
echo hi > testing.txt
cd /home/root/multicomm
/home/root/multicomm/set_pin_mux.sh
/usr/bin/python /home/root/multicomm/read_ttyO1.py &
/usr/bin/python /home/root/multicomm/read_ttyO2.py &
The script it work great when run directly from the command line.
The systemctl start multicomm.service doesn’t complain when I run it. I can even enable it and disable it. But it never works unless I run it directly. Any thoughts?
Additionally when I run the command systemctl status multicomm.service it shows me this:
multicomm.service - Run the LED pattern script
Loaded: loaded (/etc/systemd/system/multicomm.service; enabled)
Active: inactive (dead) since Sat, 01 Jan 2000 00:27:49 +0000; 12 years and 5 months ago
Process: 95 ExecStart=/home/root/multicomm/run.sh (code=exited, status=0/SUCCESS)
CGroup: name=systemd:/system/multicomm.service
Jan 01 00:27:48 beaglebone run.sh[95]: name: uart1_rxd.uart1_rxd (0x44e10980/0x980 = 0x0020), b NA, t NA
Jan 01 00:27:48 beaglebone run.sh[95]: mode: OMAP_PIN_OUTPUT | OMAP_MUX_MODE0
Jan 01 00:27:48 beaglebone run.sh[95]: signals: uart1_rxd | mmc1_sdwp | d_can1_tx | i2c1_sda | NA | pr1_uart0_rxd_mux1 | N…pio0_14
Jan 01 00:27:48 beaglebone run.sh[95]: name: uart1_txd.uart1_txd (0x44e10984/0x984 = 0x0000), b NA, t NA
Jan 01 00:27:48 beaglebone run.sh[95]: mode: OMAP_PIN_OUTPUT | OMAP_MUX_MODE0
Jan 01 00:27:48 beaglebone run.sh[95]: signals: uart1_txd | mmc2_sdwp | d_can1_rx | i2c1_scl | NA | pr1_uart0_txd_mux1 | N…pio0_15
Jan 01 00:27:48 beaglebone run.sh[95]: name: spi0_d0.uart2_txd (0x44e10954/0x954 = 0x0001), b NA, t NA
Jan 01 00:27:48 beaglebone run.sh[95]: mode: OMAP_PIN_OUTPUT | OMAP_MUX_MODE1
Jan 01 00:27:48 beaglebone run.sh[95]: signals: spi0_d0 | uart2_txd | i2c2_scl | NA | NA | NA | NA | gpio0_3
Which is the correct output given the script running, yet still nothing. No scripts running, no file appearing. Nada.
have you tried removing the ampersand from you shell script?
It’s weird in the shell script, I had a similar problem and it didn’t work for me until I provided a dump file: which is in your case could be:
/usr/bin/python /home/root/multicomm/read_ttyO1.py > /home/root/pythonlog.log
Thanks for the explanations, and the examples. I get to the same point as the last person who responded with issues. If I run the shell script from the command line, it works fine. If I start it via the systemd service, I can see evidence that it runs, because I re-direct the output of the shell script to a log file, and the file gets created, but there are no contents to the file, as if the process is hung. There is additional evidence of the process hanging because if I run “ps -e | grep python”, there is one process that is running, but I cannot tell what state the process is in. The python program I execute from the shell script is supposed to spawn 4 other processes, each running a python program. That never happens, at least not when systemd runs the shell script that is pointed to by the new service I created. If I comment out the python program, and just have the shell script echo something to the console, it executes fine, and the output shows up in what the last poster refers to as a dump-file.
I have fully specified all files in the shell commands, so I know it finds them, but now I am wondering if there is something about my python execution environment that works based on my login variables, but will not work via the systemd service? For example, when one puts in “import time”, is it possible that python finds that set of libraries based on environment parameters in my login that are not there for systemd? Maybe python is hanging looking for shared libraries?
Not sure if anyone is looking at this old thread any more, but I think I found the problem, though I have not found a solution. I was able to successfully define a service that starts a python program on reboot when the python program is a simple program that does not try to spawn sub-processes. So, its not a path problem in that my “simple” program uses python modules like PySerial, PyBBIO, time, system, and all works well. When, however, I try to start up the python process that I really want to run, which is one that uses the python library “subprocess” and tried to spawn multiple processes using the subprocess.Popen() method, the service seems to hang. I don’t know if there is some limit on which kind of process can spawn sub-processes, and whether thats allowed in systemd services or not. Based on some other posts I found, I change the Type indicator in the service to “Type=forking”, but that does not seem to have changed anything.