Comment # 10 on bug 993516 from Donald Curtis

Further investigation with this morning's boot:
--------------------------------------------------------------------
Aug 18 08:52:10 eck001 kernel: r8169 Gigabit Ethernet driver 2.3LK-NAPI loaded
Aug 18 08:52:15 eck001 systemd[1]: Started wicked network management service
daemon.
Aug 18 08:52:15 eck001 systemd[1]: Started wicked network nanny service.
Aug 18 08:52:16 eck001 kernel: r8169 0000:02:00.0 enp2s0: link down
Aug 18 08:52:16 eck001 kernel: r8169 0000:02:00.0 enp2s0: link down
Aug 18 08:52:16 eck001 kernel: IPv6: ADDRCONF(NETDEV_UP): enp2s0: link is not
ready
Aug 18 08:52:18 eck001 kernel: r8169 0000:02:00.0 enp2s0: link up
Aug 18 08:52:18 eck001 kernel: IPv6: ADDRCONF(NETDEV_CHANGE): enp2s0: link
becomes ready
Aug 18 08:52:20 eck001 avahi-daemon[885]: New relevant interface enp2s0.IPv6
for mDNS.
Aug 18 08:52:20 eck001 avahi-daemon[885]: Registering new address record for
fe80::16da:e9ff:feec:a04d on enp2s0.*.
Aug 18 08:52:22 eck001 wicked[959]: lo              up
Aug 18 08:52:22 eck001 wicked[959]: enp2s0          up
Aug 18 08:52:27 eck001 wickedd-dhcp6[953]: enp2s0: Accepting best DHCPv6 lease
offer with weight 1 from server fe80::16da:e9ff:feec:a04d
Aug 18 08:52:27 eck001 wickedd-dhcp6[953]: enp2s0: Committed DHCPv6 lease
--------------------------------------------------------------------
Time between the kernel loading the driver and link up/ready: 8 s.
Time between the kernel loading the driver and systemd starting wicked: 5 s.
Time between wicked being started and acknowledging the interface as being
"up": 7 s.
Time between wicked acknowledging that the interface as being "up" and the
DHCPv6 lease being committed: 3 s.
--------------------------------------------------------------------
Given 8 s between the kernel loading the driver and acknowledging the link as
being up/ready, the value of 8 s in config:WAIT_FOR_INTERFACES is possibly too
tight -- I'll increase it to 10 s.
The values of dhcp:DHCLIENT_WAIT_AT_BOOT and dhcp:DHCLIENT6_WAIT_AT_BOOT could
be increased to 5 s but, I can't see any reason to do so:
 - there seems to be only a requirement that WAIT_FOR_INTERFACES is at least
double the value of the DHCP client "wait at boot" times.
--------------------------------------------------------------------
One point that should maybe not be overlooked is that, the AVM FRITZ!Box 7490
DSL Router (and possibly other AVM models as well) seems to need the DHCP
clients to sign off (release) before the DHCP client shuts down.
--------------------------------------------------------------------
Does this "wicked tuning" procedure need to be documented?