TBS: TinkerOS v2.0.8 (Debian Stretch) zombify bash scripts!

[6a5h4 0]

Why TBS zombify bash scipts that are running as services (daemons) after some hours or days; the zombie itself uses no more cpu; but the parent process (if a subshell was called it is my script or if my main script gets be zombified the parent is /bin/bash) is using one complete core load; so the TBS is consuming a lot of electrical power and heating up -- SO USELESS !!! I now serving the devil by two other bash scripts as services (daemons) who are watching/monitoring CPU FEQ, CPU TEMP, RAM free, eMMC free and if there is a zombified task -- then the script kills the parent automatically if it is longer as a half minute a zombie and SIGTERM don't end this task... systemd restarts the killed service as soon as possible (mostly within 200ms) - that works fine...

Why can't systemd restart services which are using full core load? Or is it configurable in the service-file of systemd? Howto let systemd watch the services correctly? It can only restart a died service; but the child of my service gets zombified...

I tested the service scripts on 2 TBS; both did it the same way... on an odroid N2+ with Ubuntu 20.04 LTS (x64) it's running pretty fine for months (!!) - how is that possible? Is the x86 buggy or are some parallel tasks (ping ips and write $? to RAM) to much??

[Frank3_Wu 2]

Hi @6a5h4,

Can you share your system logs? And part of your script so we can duplicate and check?

e.g. in that script, do they have loops or some to let polling rate too higher and with no suspend?

 

or trying with our new OS version? since v2.0.8 is kind of older (20181023). 

https://tinker-board.asus.com/download-list.html?product=tinker-board-s

Debian latest version is v3.0.11. (First Debian 10 for Tinker Board series) or Debian 9 V2.2.9.

Thanks.