oci-reboot.sh

#!/bin/bash

res=$(nmap 127.0.0.1 -PN -p ssh | grep open)
curl -s -m 10 --retry 5 --output /dev/null https://healthcheck.io/ping/UUID/start
curl -fsS -m 10 --retry 5 --output /dev/null --data "${res}" https://healthcheck.io/ping/UUID/log
if [[ "${res}" =~ "open" ]]; then
    echo "`date` All good" >> /home/pi/ocicheck.log
    curl -s -m 10 --retry 5 --output /dev/null https://healthcheck.io/ping/UUID
    rm -f /home/pi/awaitingboot
else
    if [ -f "/home/pi/awaitingboot" ]; then
        echo "`date` Awaiting reboot" >> /home/pi/ocicheck.log
    else
        echo "`date` Connection failed" >> /home/pi/ocicheck.log
        curl -fsS -m 10 --output /dev/null --retry 5 -q --data "`date` connection failure" https://healthcheck.io/ping/UUID/log
        INSTANCE=`oci compute instance list -c ocid1.tenancy.INSTANCEID | jq .data[0]`
        INSTANCE_ID=`echo $INSTANCE | jq .id`
        INSTANCE_NAME=`echo $INSTANCE | jq ".[\"display-name\"]"`
        echo "`date` Executing reset on instance $INSTANCE_NAME" >> /home/pi/ocicheck.log;
        /usr/local/bin/oci compute instance action --instance-id ${INSTANCE_ID//\"} --action reset
        touch /home/pi/awaitingboot
    fi
    curl -s -m 10 --retry 5 --output /dev/null https://healthcheck.io/ping/UUID/fail
fi

Replace the relevant things, and your Oracle Cloud Instance will be hard-reset every time it goes down... within a minute or so...
At least, that's the idea.