dhclient -1 exits 0 when no lease found

for reference, you can make it fail faster like this in /etc/dhcp/dhclient.conf

timeout 3;
retry 3;
#reboot 10;
select-timeout 5;
initial-interval 1;

Here is the behavior on either 14.04 or 16.04:

# dhclient -1 -v virbr0
Internet Systems Consortium DHCP Client 4.2.4
Copyright 2004-2012 Internet Systems Consortium.
All rights reserved.
For info, please visit https://www.isc.org/software/dhcp/

Listening on LPF/virbr0/52:54:00:05:e1:f3
Sending on LPF/virbr0/52:54:00:05:e1:f3
Sending on Socket/fallback
DHCPDISCOVER on virbr0 to 255.255.255.255 port 67 interval 3 (xid=0xe7ef8212)
DHCPDISCOVER on virbr0 to 255.255.255.255 port 67 interval 1 (xid=0xe7ef8212)
No DHCPOFFERS received.
No working leases in persistent database - sleeping.
# ps awxfu|grep dh.*virbr
root 16796 0.0 0.0 8868 792 pts/36 S+ 05:31 0:00 | | \_ grep dh.*virbr
root 16772 0.0 0.0 10220 4368 ? Ss 05:30 0:00 \_ dhclient -1 -v virbr0
#

Not only does it not return the documented exit code, it also backgrounds itself and continues polling. So the -1 option seems to be completely broken.

I've looked at the source package, and this is due to debian/patches/onetry_retry_after_initial_success. The addition of this patch isn't documented in the changelog, there are no referenced bugs, and the patch description doesn't make sense to me; assigning to stgraber, who is listed as the author of this patch.

According to the bzr history in lp:ubuntu/wily/isc-dhcp, this patch is related to bug #974284:

committer: Stéphane Graber <email address hidden>
branch nick: isc-dhcp
timestamp: Wed 2012-09-12 17:30:23 -0400
message:
  Set onetry = 0 early enough to actually work. (LP: #974284)

And Scott, you filed that bug; it seems the current behavior is exactly what you requested?

Steve,
Good sleuthing.

Bug 974284 has good info in it.

This is definitely tricky.
I think we have the following cases to consider and may be at odds.
a.) dhclient -1 should exit failure as documented.
b.) we need a reliable point in boot where all static networking is configured.
c.) Ill timed reboot (or recovery from lab power failure)
    a system is configured for dhcp is rebooted while the dhcp server is unavailable.
    It seems better for this system to keep trying to get an IP address as otherwise
    the system will be offline forever.
d.) Any dhcp server downtime and failed lease renewal.
    a system is configured for dhcp and initially succeeds, but while re-trying a lease
    the dhcp server is down for scheduled maintenance. A failure to release here
    results in system offline if dhclient takes down the interface.
    This is mentioned at https://bugs.launchpad.net/ubuntu/+source/ifupdown/+bug/838968/comments/11