I have installed medusa from dietpi-software and it runs fine.
But if I
# systemctl status medusa
I get
● medusa.service - Medusa Daemon
Loaded: loaded (/etc/systemd/system/medusa.service; disabled; vendor preset: enabled)
Active: inactive (dead) since Sat 2019-05-25 15:32:17 NZST; 1min 38s ago
Process: 23645 ExecStart=/usr/bin/python /mnt/dietpi_userdata/medusa/start.py -q --daemon --nolaunch --datadir=/mnt/dietpi_userdata/medusa (code=exited, status=0
/SUCCESS)
Main PID: 23645 (code=exited, status=0/SUCCESS)
Memory: 707.6M
CPU: 13.363s
CGroup: /system.slice/medusa.service
└─2058 /usr/bin/python /mnt/dietpi_userdata/medusa/start.py -q --daemon --nolaunch --datadir=/mnt/dietpi_userdata/medusa
May 25 15:32:05 Odroid systemd[1]: Started Medusa Daemon.
It is running fine but somehow systemd doesn't know it is running.
Any idea why this is?
John
Medusa question Topic is solved
Re: Medusa question
@johnvick
Just verified it. On service start, first some compiling steps are done that take some seconds with 100% CPU usage. Afterwards the process exits and a new one is spawned. Since systemd just sees the main process exiting but cannot know about the respawn, it shows it as exited/dead.
Will check the official Medusa instructions, perhaps they suggest a systemd unit that detects the process state correctly, e.g. with Type=forking instead of Type=simple.
Fixed with: https://github.com/MichaIng/DietPi/comm ... ebe5e0d131
- We used used the "--daemon" option that should not be used within systemd, besides you use a different type with then PID file required.
However a minor one, so I think we do not need to actively patch it for existing installs?
Just verified it. On service start, first some compiling steps are done that take some seconds with 100% CPU usage. Afterwards the process exits and a new one is spawned. Since systemd just sees the main process exiting but cannot know about the respawn, it shows it as exited/dead.
Will check the official Medusa instructions, perhaps they suggest a systemd unit that detects the process state correctly, e.g. with Type=forking instead of Type=simple.
Fixed with: https://github.com/MichaIng/DietPi/comm ... ebe5e0d131
- We used used the "--daemon" option that should not be used within systemd, besides you use a different type with then PID file required.
However a minor one, so I think we do not need to actively patch it for existing installs?
Re: Medusa question
@MichaIng
Thanks for tracking this down - it had been puzzling me when i was trying to get Medusa to run under Python 3 as it looked like Medusa had crashed when it hadn't. Anyway all good now with Python 3.5.3 (the Medusa team are encouraging people to use Python 3 but is not that easy).
John
Thanks for tracking this down - it had been puzzling me when i was trying to get Medusa to run under Python 3 as it looked like Medusa had crashed when it hadn't. Anyway all good now with Python 3.5.3 (the Medusa team are encouraging people to use Python 3 but is not that easy).
John