using the provided weewx.service on an RPi, ‘sudo systemctl stop weewx’ sends
SIGKILL instantly (twice), which is a very unclean termination of weewx and
causes me problems. has anyone encountered this and has a proper fix?
so far as i can tell from the manuals (particularly ‘systemd.kill' and
SIGKILL is unclean behavior, but it should not cause any harm. The only
state worth worrying about is the database and, there, all writes are done
as transactions. If the SIGKILL happens in the middle of one, it will not
be committed.
Of course, that's the theory!
On Fri, Apr 2, 2021 at 7:36 AM G
some of my services do have state that i need to save on termination, so it is
important to me
> On 3 Apr 2021, at 1:42 am, Tom Keffer wrote:
>
> SIGKILL is unclean behavior, but it should not cause any harm. The only state
> worth worrying about is the database and, there, all writes are done
KILL is definitely not desirable. i'm debugging this now on centos with
systemd, weewx, and rtl_433.
for example, the sdr driver spawns the rtl_433 command, and if you kill
weewx, you end up with a rogue process, and the next time you start weewx,
weewx will fail.
as another example, if you a