In order to achieve boot-persistence you have a couple of options:
first you need to place your config file in persistent storage (/mnt/flash, /persist/sys are a few examples – USB and SSD would also work)
then you need to copy your config file to the correct location on-boot. You can do this by either using:
/mnt/flash/rc.eos – this can be a bash script (or a Python script, etc.) which copies your file from the persistent storage to the correct location
use an on-boot event handler which does the same thing
you can use the same mechanism to execute any restart/reconfigure commands after copying the file to the correct location
The difference between the two option from above is the fact that rc.eos runs very early in the boot process (before any agents or Sysdb are started, whereas the on-boot event handler runs later). Depending on what you are trying to achieve one or the other are a better option.
Got it. Thank you for the explanation. Regarding the specific case of the ganglia daemon, I had to add a 60 delay in the event-handler to prevent a ”unable to create udp socket” from gmond itself. Maybe the full network stack is not ready on the on-boot event ?
Correct – the delay should fix it.