Network Working Group B. Liu
Internet Draft Huawei Technologies
Intended status: Informational R. Bonica
Expires: December 20, 2014 Juniper Networks
S. Jiang
Huawei Technologies
X. Gong
W. Wang
BUPT University
June 18, 2014
DHCPv6/SLAAC Address Configuration Interaction Problem Statement
draft-ietf-v6ops-dhcpv6-slaac-problem-01.txt
Status of this Memo
This Internet-Draft is submitted in full conformance with the
provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF). Note that other groups may also distribute working
documents as Internet-Drafts. The list of current Internet-Drafts is
at http://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months
and may be updated, replaced, or obsoleted by other documents at any
time. It is inappropriate to use Internet-Drafts as reference
material or to cite them other than as "work in progress."
This Internet-Draft will expire on December 18, 2014.
Copyright Notice
Copyright (c) 2011 IETF Trust and the persons identified as the
document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents
(http://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents
carefully, as they describe your rights and restrictions with respect
to this document. Code Components extracted from this document must
include Simplified BSD License text as described in Section 4.e of
the Trust Legal Provisions and are provided without warranty as
described in the Simplified BSD License.
Liu, et al. Expires December 20 2014 [Page 1]
Internet-Draft draft-ietf-v6ops-dhcpv6-slaac-problem June 2014
Abstract
This document analyzes the DHCPv6/SLAAC interaction issue on host.
More specifically, the interaction is regarding with the A, M, and O
flags which are defined in ND protocol. Test results identify that
current implementations in operating systems have varied on
interpreting the flags. The variation might cause some operational
issues as described in the document.
Table of Contents
1. Introduction ................................................. 3
2. Host Behavior Definition in Standards ........................ 3
2.1. A (Autonomous) Flag ..................................... 4
2.2. M (Managed) Flag ........................................ 4
2.3. O (Otherconfig) Flag .................................... 4
3. Problems Statement ........................................... 5
3.1. Host Behavior Ambiguity ................................. 5
3.2. Operational Problems Implication ........................ 6
3.2.1. Renumbering ........................................ 6
3.2.2. Cold Start Problem ................................. 6
3.2.3. Specific Management Patterns ....................... 7
4. Conclusions .................................................. 7
5. Security Considerations ...................................... 7
6. IANA Considerations .......................................... 7
7. References ................................................... 7
7.1. Normative References .................................... 7
7.2. Informative References .................................. 8
8. Acknowledgments .............................................. 8
Appendix A. Test Results of Host Behavior ....................... 9
A.1 Detailed Test Results .................................... 9
A.1.1 Host Initial Behavior .............................. 10
A.1.2 Host Behavior in Flags Transition .................. 10
A.2 Observations from the Test .............................. 11
Liu, et al. Expires December 20, 2014 [Page 2]
Internet-Draft draft-ietf-v6ops-dhcpv6-slaac-problem June 2014
1. Introduction
In IPv6, both of the DHCPv6 [RFC3315] and Neighbor Discovery [RFC4861]
protocols could be utilized for automatic IP address configuration
for the hosts. They are known as stateful address auto-configuration
and stateless address auto-configuration (SLAAC, [RFC4862]).
Sometimes the two address configuration methods might be both
available in one network.
In ND protocol, there is an M (Managed) flag defined in RA message,
indicating the hosts whether there is DHCPv6 service in the network
or not. And there is an O (OtherConfig) flag, if set, indicating
configure information other than addresses (e.g. DNS, Route .etc) is
available through DHCPv6 configuration. Moreover, there's another A
(Autonomous) flag defined in ND, indicating the hosts to do SLAAC,
may also influent the behavior of hosts.
So with these flags, the two address configuration mechanisms are
somehow correlated. But for some reasons, the ND protocol didn't
define the flags as prescriptive but only advisory. This ambiguous
definition might vary the behavior of interpreting the flags by
different hosts. This would add additional complexity for both the
hosts and the network management.
This draft reviews the standard definition of the above mentioned
flags, and identifies the potential ambiguous behavior of
interpreting these flags. And then analyzes what operational problems
might be caused by the ambiguous behavior.
In the appendix, detailed test results of several major desktop
operating systems' behavior of interpreting the flags are provided.
According to the test results, we can see the ambiguity problem is
actually happening in current implementations.
2. Host Behavior Definition in Standards
In this section, we analyzed A, M and O flags definition.
Please note that, A flag has no direct relationship with DHCPv6, but
it is somewhat correlated with M and O flags.
Liu, et al. Expires December 20, 2014 [Page 3]
Internet-Draft draft-ietf-v6ops-dhcpv6-slaac-problem June 2014
2.1. A (Autonomous) Flag
In ND Prefix Information Option, the autonomous address-configuration
flag (A flag)is used to indicate whether this prefix can be used for
SLAAC.
For the host behavior, there is an explicit rule in the SLAAC
specification [RFC4862]: "If the Autonomous flag is not set, silently
ignore the Prefix Information option."
But when A flag is set, the SLAAC protocol didn't provide a
prescriptive definition. (However, it is quite obvious that host
should do SLAAC when A flag is set.)
2.2. M (Managed) Flag
In earlier SLAAC specification [RFC2462], the host behavior of
interpreting M flag is as below:
"On receipt of a valid Router Advertisement, a host copies the value
of the advertisement's M bit into ManagedFlag. If the value of
ManagedFlag changes from FALSE to TRUE, and the host is not already
running the stateful address autoconfiguration protocol, the host
should invoke the stateful address auto-configuration protocol,
requesting both address information and other information. If the
value of the ManagedFlag changes from TRUE to FALSE, the host should
continue running the stateful address auto-configuration, i.e., the
change in the value of the ManagedFlag has no effect. If the value
of the flag stays unchanged, no special action takes place. In
particular, a host MUST NOT reinvoke stateful address configuration
if it is already participating in the stateful protocol as a result
of an earlier advertisement."
But in the updated SLAAC specification [RFC4862], the relative
description was removed, the reason was "considering the maturity of
implementations and operational experiences. ManagedFlag and
OtherConfigFlag were removed accordingly. (Note that this change does
not mean the use of these flags is deprecated.)"
2.3. O (Otherconfig) Flag
The situation of O flag is similar with above mentioned M flag. In
earlier SLAAC [RFC2462], the host behavior is clear:
"If the value of OtherConfigFlag changes from FALSE to TRUE, the host
should invoke the stateful autoconfiguration protocol, requesting
information (excluding addresses if ManagedFlag is set to FALSE). If
Liu, et al. Expires December 20, 2014 [Page 4]
Internet-Draft draft-ietf-v6ops-dhcpv6-slaac-problem June 2014
the value of the OtherConfigFlag changes from TRUE to FALSE, the host
should continue running the stateful address autoconfiguration
protocol, i.e., the change in the value of OtherConfigFlag has no
effect. If the value of the flag stays unchanged, no special action
takes place. In particular, a host MUST NOT reinvoke stateful
configuration if it is already participating in the stateful protocol
as a result of an earlier advertisement."
And there's another description of the relationship of M and O flags
in [RFC2462]:
"In addition, when the value of the ManagedFlag is TRUE, the value of
OtherConfigFlag is implicitely TRUE as well. It is not a valid
configuration for a host to use stateful address autoconfiguration to
request addresses only, without also accepting other configuration
information."
3. Problems Statement
3.1. Host Behavior Ambiguity
The main problem is standard definition ambiguity which means, on
interpreting the same messages, different hosts might behave
differently. Thus it could be un-controlled or un-predictable for
administrators on some operations. The ambiguity is summarized as the
following aspects.
#1 Dependency between DHCPv6 and RA
In standards, behavior of DHCPv6 and Neighbor Discovery protocols is
specified respectively. But it is not clear that whether there should
be any dependency between them.
More specifically, is RA (with M=1) required to trigger DHCPv6? If
there are no RAs at all, should hosts initiate DHCPv6 by themselves?
#2 Advisory or Prescriptive
Some platforms interpret the flags as advisory while others might
interpret them prescriptive. Especially when flags are in transition,
e.g. the host is already SLAAC-configured, then M flag changes from
FALSE to TRUE, it is not clear whether the host should start DHCPv6
or not; or vise versa, the host is already both SLAAC/DHCPv6
configured, then M flag change from TRUE to FALSE, it is also not
clear whether the host should turn DHCPv6 off or not.
#3 "Address Configuring Method" and "Address Lifetime"
Liu, et al. Expires December 20, 2014 [Page 5]
Internet-Draft draft-ietf-v6ops-dhcpv6-slaac-problem June 2014
When one address configuration method is off, that is, the A flag or
M flag changes from TRUE to FALSE, it is not clear whether the host
should immediately release the corresponding address(es) or just
retain it(them) until expired.
#4 Dependencies between the flags
The semantics of the flags seems not totally independent, but the
standards didn't clearly clarify it. For example, when both M and O
flags are TRUE, it is not clear whether the host should initiate one
stateful DHCPv6 session to get both address and info-configuration or
initiate two independent sessions of which one is dedicated for
address provisioning and the other is for information provision. When
A and M flags are FALSE and O flag is TRUE, it is not clear whether
the host should initiate a stand-alone stateless DHCPv6 session.
3.2. Operational Problems Implication
According to the abovementioned host behavior ambiguity, there might
be operational issues as the following.
3.2.1. Renumbering
During IPv6 renumbering, the SLAAC-configured hosts can reconfigure
IP addresses by receiving ND Router Advertisement (RA) messages
containing new prefix information. The DHCPv6-configured hosts can
reconfigure addresses by initialing RENEW sessions when the current
addresses' lease time is expired or receiving the reconfiguration
messages initialed by the DHCPv6 servers.
The above mechanisms have an implicit assumption that SLAAC-
configured hosts will remain SLAAC while DHCPv6-managed hosts will
remain DHCPv6-managed. But in some situations, SLAAC-configured hosts
might need to switch to DHCPv6-managed, or vice versa. In [RFC6879],
it described several renumbering scenarios in enterprise network for
this requirement; for example, the network may split, merge, relocate
or reorganize. But due to current implementations, this requirement
is not applicable and has been identified as a gap in [RFC7010].
3.2.2. Cold Start Problem
If all nodes, or many nodes, restart at the same time after a power
cut, the results might not consistent.
Liu, et al. Expires December 20, 2014 [Page 6]
Internet-Draft draft-ietf-v6ops-dhcpv6-slaac-problem June 2014
3.2.3. Specific Management Patterns
Since the host behavior of address configuration is somehow un-
controlled by the network side, it might cause gaps to the networks
that need some specific management patterns. Examples are:
- the hosts have been SLAAC-configured, then the network need the
hosts to do DHCPv6 simultaneously (e.g. for multihoming).
- the network wants the hosts to do stateless DHCPV6-only; for
example, the hosts are configured with self-generated addresses (e.g.
ULA), and they also need to contact the DHCPv6 server for info-
configuration.
4. Conclusions
- The host behavior of SLAAC/DHCPv6 interaction is ambiguous in
standard.
- Varied behavior of implementations has been observed. In [RFC4862]
it is said "Removed the text regarding the M and O flags, considering
the maturity of implementations and operational experiences." This
consideration intended to remain the ambiguity. But in the
perspective of operation, ambiguity normally is problematic.
- It is foreseeable that the un-uniformed host behavior can cause
operational problems.
5. Security Considerations
No more security considerations than the Neighbor Discovery protocol
[RFC4861].
6. IANA Considerations
None.
7. References
7.1. Normative References
[RFC4861] Narten, T., Nordmark, E., Simpson, W., and H. Soliman,
"Neighbor Discovery for IP version 6 (IPv6)", RFC 4861,
September 2007.
[RFC4862] Thomson, S., Narten, T., and T. Jinmei, "IPv6 Stateless
Address Autoconfiguration", RFC 4862, September 2007.
Liu, et al. Expires December 20, 2014 [Page 7]
Internet-Draft draft-ietf-v6ops-dhcpv6-slaac-problem June 2014
7.2. Informative References
[RFC2462] Thomson, S. and T. Narten, "IPv6 Stateless Address
Autoconfiguration", RFC 2462, December 1998.
[RFC3315] R. Droms, Bound, J., Volz, B., Lemon, T., Perkins, C., and
M. Carney, "Dynamic Host Configuration Protocol for IPv6
(DHCPv6)", RFC 3315, July 2003.
[RFC3736] Droms, R., "Stateless Dynamic Host Configuration Protocol
(DHCP) Service for IPv6", RFC 3736, April 2004.
[RFC5887] Carpenter, B., Atkinson, R., and H. Flinck, "Renumbering
Still Needs Work", RFC 5887, May 2010.
[RFC7010] Liu, B., Jiang, S., Carpenter, B., Venaas, S., and W.
George, "IPv6 Site Renumbering Gap Analysis", RFC 7010,
September 2013.
[RFC6879] Jiang, S., Liu, B., and B. Carpenter, "IPv6 Enterprise
Network Renumbering Scenarios, Considerations, and Methods",
RFC 6879, February 2013.
8. Acknowledgments
The test was done by our research partner BNRC-BUPT (Broad Network
Research Centre in Beijing University of Posts and
Telecommunications). Thanks for the hard efficient work of student
Xudong Shi and Longyun Yuan.
Valuable comments were received from Sheng Jiang, Brian E Carpenter,
Ron Atkinson, Mikael Abrahamsson, Tatuya Jinmei, Mark Andrews and
Mark Smith to improve the draft.
This document was prepared using 2-Word-v2.0.template.dot.
Liu, et al. Expires December 20, 2014 [Page 8]
Internet-Draft draft-ietf-v6ops-dhcpv6-slaac-problem June 2014
Appendix A. Test Results of Host Behavior
We did tests of the behavior of interpreting the flags by current
mainstream desktop/mobile operating systems as the following.
A.1 Detailed Test Results
/-----\
+---------+ // \\
| DHCPv6 | | Router |
| server | \\ //
+----+----+ \--+--/
| |
| |
| |
----+--+----------+----------+---+-----
| | |
| | |
| | |
+----+---+ +----+---+ +----+---+
| | | | | |
| Host1 | | Host2 | | Host3 |
+--------+ +--------+ +--------+
Figure 1 Test Environment
The 5 elements were all created in Vmware in one computer, for ease
of operation.
- Router quagga 0.99-19 soft router installed on Ubuntu 11.04
virtual host
- DHCPv6 Server: dibbler-server installed on Ubuntu 11.04 virtual
host
- Host A Window 7 Virtual Host
- Host B Ubuntu 12.10 Virtual Host
- Host C Mac OS X v10.7 Virtual Host
Another test was done dedicated for the mobile phone operating
systems. The environment is similar (not in VMware, all are real PC
and mobile phones):
- Router quagga 0.99-17 soft router installed on Ubuntu 12.10
- DHCPv6 Server: dibbler-server installed on Ubuntu 12.10
Liu, et al. Expires December 20, 2014 [Page 9]
Internet-Draft draft-ietf-v6ops-dhcpv6-slaac-problem June 2014
- Host D Android 4.0.4 (kernel: 3.0.16-gfa98030; device: HTC
Incredible S)
- Host E IOS 6.1.3 (model: iPod Touch 4)
(Note: The tested Android version didn't support DHCPv6, so the
following results don't include Android.)
A.1.1 Host Initial Behavior
When hosts are not configured yet, we tested their behavior when
receiving different A/M/O combinations. The results are as the
following:
o Window 7/Apple IOS
- A=0&M=O&O=0, non-config
- A=1&M=0&O=0, SLAAC only
- A=1&M=0&O=1, SLAAC + Stateless DHCPv6
- A=1&M=1&O=0, SLAAC + DHCPv6
- A=1&M=1&O=1, SLAAC + DHCPv6
- A=0&M=1&O=0, DHCPv6 only (A=0 or Non-PIO)
- A=0&M=1&O=1, DHCPv6 only (A=0 or Non-PIO)
- A=0&M=0&O=1, Stateless DHCPv6 only
o Linux/MAC OS X
- A=0&M=O&O=0, non-config
- A=1&M=0&O=0, SLAAC only
- A=1&M=0&O=1, SLAAC + Stateless DHCPv6
- A=1&M=1&O=0, SLAAC + DHCPv6
- A=1&M=1&O=1, SLAAC + DHCPv6
- A=0&M=1&O=0, DHCPv6 only (A=0 or Non-PIO)
- A=0&M=1&O=1, DHCPv6 only (A=0 or Non-PIO)
- A=0&M=0&O=1, non-config
As showed above, Linux and MAC OSX acted the same way, but are
different from Windows 7 and Apple IOS. The only difference is when
A=0&M=0&O=1, Windows 7/Apple IOS did stateless DHCPv6 while Linux/MAC
OSX did nothing.
A.1.2 Host Behavior in Flags Transition
o SLAAC-only host receiving M=1
- Window 7 would initiate DHCPv6
- Linux/MAC/IOS would keep SLAAC and don't initiate DHCPv6 unless
SLAAC is expired and no continuous RAs
o DHCPv6-only host receiving A=1
Liu, et al. Expires December 20, 2014 [Page 10]
Internet-Draft draft-ietf-v6ops-dhcpv6-slaac-problem June 2014
- They all do SLAAC
o Stateless DHCPv6-configured host receiving M=1 (while keeping O=1)
- Window 7 would initiate stateful DHCPv6, configuring address as
well as re-configuring other information
- Linux/MAC/IOS no action
o Statefull DHCPv6-configured host receiving M=0 (while keeping O=1)
- Window 7 would release all DHCPv6 configurations including
address and other information, and initiate stateless DHCPv6
- Linux/MAC/IOS no action
A.2 Observations from the Test
o A flag
A flag is a switch to control whether to do SLAAC, and it is
independent with M and O flags, in another word, A is independent
with DHCPv6.
At the non-SLAAC-configured state (either non-configured or DHCPv6-
configured only), all the operating systems acted the same way in
interpreting A flag. If A flag is TRUE, they all configure SLAAC, it
is obvious and reasonable.
o M flag
M is a key flag to interact ND and DHCPv6, but the host behaviors on
M flag were quite different.
At the initialing state, some operating systems would start DHCPv6
only if receiving an RA message with M flag set while some would
initially start DHCPv6 if RAs are absent. This result reflects the
ambiguity problem of #1 Dependency between DHCPv6 and RA in above
text.
When the hosts are SLAAC-configured, and then the M flag changes from
FALSE to TRUE, some operating systems would initiate DHCPv6 while
some would not. This reflects the problem #2 Advisory or Prescriptive.
o O flag
In the test, when M flag is set, the O flag is implicitly set as well;
in another word, the hosts would not initial stateful DHCPv6 and
stateless DHCPv6 respectively. This is a reasonable behavior.
Liu, et al. Expires December 20, 2014 [Page 11]
Internet-Draft draft-ietf-v6ops-dhcpv6-slaac-problem June 2014
But the O flag is not independent from A flag in some operating
systems, which won't initiate stateless DHCPv6 when A flag is FALSE.
That is to say, it is not applicable to have a "stateless DHCPv6
only" configuration state for some operating systems; it is also not
applicable for these operating systems to switch between stateful
DHCPv6 and stateless DHCPv6 (according to O flag changing from TRUE
to FALSE or vice versa). This reflects the problem #4 Dependencies
between the flags.
Liu, et al. Expires December 20, 2014 [Page 12]
Internet-Draft draft-ietf-v6ops-dhcpv6-slaac-problem June 2014
Authors' Addresses
Bing Liu
Q14-4-A Building
Huawei Technologies Co., Ltd
Zhong-Guan-Cun Environment Protection Park, No.156 Beiqing Rd.
Hai-Dian District, Beijing
P.R. China
Email: leo.liubing@huawei.com
Ron Bonica
Juniper Networks
Sterling, Virginia 20164
USA
Email: rbonica@juniper.net
Xiangyang Gong
No.3 Teaching Building
Beijing University of Posts and Telecommunications (BUPT)
No.10 Xi-Tu-Cheng Rd.
Hai-Dian District, Beijing
P.R. China
Email: xygong@bupt.edu.cn
Wendong Wang
No.3 Teaching Building
Beijing University of Posts and Telecommunications (BUPT)
No.10 Xi-Tu-Cheng Rd.
Hai-Dian District, Beijing
P.R. China
Email: wdwang@bupt.edu.cn