Restart long-running systemd services

[arnaldo2792]

Issue number:
N / A

Description of changes:

packages: restart long-running systemd services

This adds the 'Restart' and 'RestartSec' directives to the following
services since they must be running even if they exit or are killed by
the OS:

- acpid
- vmtoolsd
- apiserver

Testing done:
I executed this script from the admin container to confirm that the services were restarted after the main process was killed:

SYSTEMCTL_CMD="chroot /.bottlerocket/rootfs /usr/bin/systemctl"
KILL_CMD="chroot /.bottlerocket/rootfs /usr/bin/kill"
SERVICES="acpid  apiserver"

get_pid() {
    local service
    service="${1}"
    ${SYSTEMCTL_CMD} show "${service}" | grep -E '^ExecMainPID' | awk ' { split($0,a,"="); print a[2] } '
}

for s in ${SERVICES}; do
    OLD_PID=$(get_pid "${s}")
    echo "OLD PID for ${s} is ${OLD_PID}"
    ${KILL_CMD} -9 ${OLD_PID}
    # Sleep a little longer than it takes for the service to restart
    sleep 7
    NEW_PID=$(get_pid "${s}")
    echo "NEW PID for ${s} is ${NEW_PID}"
done

Output:

OLD PID for acpid is 2291
NEW PID for acpid is 2768
OLD PID for apiserver is 2302
NEW PID for apiserver is 2779

Terms of contribution:

By submitting this pull request, I agree that this contribution is dual-licensed under the terms of both the Apache License, version 2.0, and the MIT license.

[jpculp]

If we're going to have it always restart, we might want to set some "start limits" to avoid restarting too many times too quickly if something is amiss.

[arnaldo2792]

Forced push includes:

• Remove changes in chronyd since we need a different approach for it, see Configure chronyd to adjust the time smoothly  #2170
• Add RestartSec=5 to all the modified services