This is a blocker issue discovered by our CI team while we are migrating our CI agent instances to use more up-to-date OECore.

http://cgit.openembedded.org/openembedded-core/tree/meta/recipes-core/initscripts/initscripts-1.0/sendsigs and in fact nearly all functions/helpers are installed directly under ${D}${sysconfdir}/init.d with the consequence of them ending up executed by the service command run with --status-all option. As the sendsigs unconditionally sends -15 and -9 to all daemons this brings down the system. Really not the behaviour expected when running a simple status query.
Simple fix is to patch in the sendsigs to the block list of service https://salsa.debian.org/debian/init-system-helpers/-/blob/master/script/service#L78 The root issue however is the dangerous installation location of those helper functions.
I was amazed that this feature has apparently been there starting from gatesgarth and is still there. Sendsigs is the most fatal but there are many others not intended to be run while system is running and may cause all kind of weird issues. Maybe everyone else is using systemd ...