Request for Comments: 3847 L. Ginsberg
Category: Informational Cisco Systems
July 2004
Restart Signaling for
Intermediate System to Intermediate System (IS-IS)
Status of this Memo
This memo provides information for the Internet community. It does
not specify an Internet standard of any kind. Distribution of this
memo is unlimited.
Copyright Notice
Copyright (C) The Internet Society (2004). All Rights Reserved.
Abstract
This document describes a mechanism for a restarting router to signal
to its neighbors that it is restarting, allowing them to reestablish
their adjacencies without cycling through the down state, while still
correctly initiating database synchronization.
This document additionally describes a mechanism for a restarting
router to determine when it has achieved LSP database synchronization
with its neighbors and a mechanism to optimize LSP database
synchronization, while minimizing transient routing disruption when a
router starts.
Table of Contents
1. Conventions used in this Document. . . . . . . . . . . . . . . 2
2. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
3. Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.1. Timers . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.2. Restart TLV. . . . . . . . . . . . . . . . . . . . . . . 5
3.2.1. Use of RR and RA Bits. . . . . . . . . . . . . . 6
3.2.2. Use of SA Bit. . . . . . . . . . . . . . . . . . 7
3.3. Adjacency (re)Acquisition. . . . . . . . . . . . . . . . 8
3.3.1. Adjacency Reacquisition During Restart . . . . . 8
3.3.2. Adjacency Acquisition During Start . . . . . . . 10
3.3.3. Multiple Levels. . . . . . . . . . . . . . . . . 12
3.4. Database Synchronization . . . . . . . . . . . . . . . . 12
3.4.1. LSP Generation and Flooding and SPF Computation. 13
3.4.1.1. Restarting. . . . . . . . . . . . . . . 13
3.4.1.2. Starting. . . . . . . . . . . . . . . . 15
4. State Tables . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.1. Running Router . . . . . . . . . . . . . . . . . . . . . 16
4.2. Restarting Router. . . . . . . . . . . . . . . . . . . . 17
4.3. Starting Router. . . . . . . . . . . . . . . . . . . . . 18
5. Security Considerations. . . . . . . . . . . . . . . . . . . . 18
6. IANA Considerations. . . . . . . . . . . . . . . . . . . . . . 19
7. Normative References . . . . . . . . . . . . . . . . . . . . . 19
8. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 20
9. Authors’ Addresses . . . . . . . . . . . . . . . . . . . . . . 20
10. Full Copyright Statement . . . . . . . . . . . . . . . . . . . 21
1. Conventions used in this Document
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in BCP 14, RFC-2119 [3].
If the control and forwarding functions in a router can be maintained
independently, it is possible for the forwarding function state to be
maintained across a resumption of control function operations. This
functionality is assumed when the terms "restart/restarting" are used
in this document.
The terms "start/starting" are used to refer to a router in which the
control function has either commenced operations for the first time
or has resumed operations but the forwarding functions have not been
maintained in a prior state.
The terms "(re)start/(re)starting" are used when the text is
applicable to both a "starting" and a "restarting" router.
2. Overview
The Intermediate System to Intermediate System (IS-IS) routing
protocol [RFC 1195, ISO/IEC 10589] is a link state intra-domain
routing protocol. Normally, when an IS-IS router is restarted,
temporary disruption of routing occurs due to events in both the
restarting router and the neighbors of the restarting router.
The router which has been restarted computes its own routes before
achieving database synchronization with its neighbors. The results
of this computation are likely to be non-convergent with the routes
computed by other routers in the area/domain.
Neighbors of the restarting router detect the restart event and cycle
their adjacencies with the restarting router through the down state.
The cycling of the adjacency state causes the neighbors to regenerate
their LSPs describing the adjacency concerned. This in turn causes a
temporary disruption of routes passing through the restarting router.
In certain scenarios, the temporary disruption of the routes is
highly undesirable. This document describes mechanisms to avoid or
minimize the disruption due to both of these causes.
When an adjacency is reinitialized as a result of a neighbor
restarting, a router does three things:
1. It causes its own LSP(s) to be regenerated, thus triggering SPF
runs throughout the area (or in the case of Level 2, throughout
the domain).
2. It sets SRMflags on its own LSP database on the adjacency
concerned.
3. In the case of a Point-to-Point link, it transmits a (set of)
CSNP(s) over the adjacency.
In the case of a restarting router process, the first of these is
highly undesirable, but the second is essential in order to ensure
synchronization of the LSP database.
The third action above minimizes the number of LSPs which must be
exchanged and, if made reliable, provides a means of determining when
the LSP databases of the neighboring routers have been synchronized.
This is desirable whether the router is being restarted or not (so
that the overload bit can be cleared in the router’s own LSP, for
example).
This document describes a mechanism for a restarting router to signal
that it is restarting to its neighbors, and allow them to reestablish
their adjacencies without cycling through the down state, while still
correctly initiating database synchronization.
This document additionally describes a mechanism for a restarting
router to determine when it has achieved LSP database synchronization
with its neighbors and a mechanism to optimize LSP database
synchronization and minimize transient routing disruption when a
router starts.
It is assumed that the three-way handshake [4] is being used on
Point-to-Point circuits.
3. Approach
3.1. Timers
Three additional timers, T1, T2, and T3 are required to support the
functionality defined in this document.
An instance of the timer T1 is maintained per interface, and
indicates the time after which an unacknowledged (re)start attempt
will be repeated. A typical value might be 3 seconds.
An instance of the timer T2 is maintained for each LSP database
present in the system, i.e., for a Level1/2 system, there will be an
instance of the timer T2 for Level 1 and an instance for Level 2.
This is the maximum time that the system will wait for LSPDB
synchronization. A typical value might be 60 seconds.
A single instance of the timer T3 is maintained for the entire
system. It indicates the time after which the router will declare
that it has failed to achieve database synchronization (by setting
the overload bit in its own LSP). This is initialized to 65535
seconds, but is set to the minimum of the remaining times of received
IIHs containing a restart TLV with the RA set and an indication that
the neighbor has an adjacency in the "UP" state to the restarting
router.
NOTE: The timer T3 is only used by a restarting router.
3.2. Restart TLV
A new TLV is defined to be included in IIH PDUs. The presence of
this TLV indicates that the sender supports the functionality defined
in this document and it carries flags that are used to convey
information during a (re)start. All IIHs transmitted by a router
that supports this capability MUST include this TLV.
Type 211
Length # of octets in the value field (1 to (3 + ID Length))
Value
No. of octets
+-----------------------+
| Flags | 1
+-----------------------+
| Remaining Time | 2
+-----------------------+
| Restarting Neighbor ID| ID Length
+-----------------------+
Flags (1 octet)
0 1 2 3 4 5 6 7
+--+--+--+--+--+--+--+--+
| Reserved |SA|RA|RR|
+--+--+--+--+--+--+--+--+
RR - Restart Request
RA - Restart Acknowledgement
SA - Suppress adjacency advertisement
(Note: Remaining fields are required when the RA bit is set)
Remaining Time (2 octets)
Remaining holding time (in seconds)
Restarting Neighbor System ID (ID Length octets)
The system ID of the neighbor to which an RA refers. Note:
Implementations based on earlier versions of this document may not
include this field in the TLV when the RA is set. In this case, a
router which is expecting an RA on a LAN circuit SHOULD assume that
the acknowledgement is directed at the local system.
3.2.1. Use of RR and RA Bits
The RR bit is used by a (re)starting router to signal to its
neighbors that a (re)start is in progress, that an existing adjacency
SHOULD be maintained even under circumstances when the normal
operation of the adjacency state machine would require the adjacency
to be reinitialized, to request a set of CSNPs, and to request
setting of the SRMflags.
The RA bit is sent by the neighbor of a (re)starting router to
acknowledge the receipt of a restart TLV with the RR bit set.
When the neighbor of a (re)starting router receives an IIH with the
restart TLV having the RR bit set, if there exists on this interface
an adjacency in state "UP" with the same System ID, and in the case
of a LAN circuit, with the same source LAN address, then,
irrespective of the other contents of the "Intermediate System
Neighbors" option (LAN circuits) or the "Point-to-Point Three-Way
Adjacency" option (Point-to-Point circuits):
a) the state of the adjacency is not changed. If this is the first
IIH with the RR bit set that this system has received associated
with this adjacency, then the adjacency is marked as being in
"Restart mode" and the adjacency holding time is refreshed -
otherwise the holding time is not refreshed. The "remaining time"
transmitted according to (b) below MUST reflect the actual time
after which the adjacency will now expire. Receipt of a normal
IIH with the RR bit reset will clear the "Restart mode" state.
This procedure allows the restarting router to cause the neighbor
to maintain the adjacency long enough for restart to successfully
complete while also preventing repetitive restarts from
maintaining an adjacency indefinitely. Whether an adjacency is
marked as being in "Restart mode" or not has no effect on
adjacency state transitions.
b) immediately (i.e., without waiting for any currently running timer
interval to expire, but with a small random delay of a few 10s of
milliseconds on LANs to avoid "storms") transmit over the
corresponding interface an IIH including the restart TLV with the
RR bit clear and the RA bit set, in the case of Point-to-Point
adjacencies having updated the "Point-to-Point Three-Way
Adjacency" option to reflect any new values received from the
(re)starting router. (This allows a restarting router to quickly
acquire the correct information to place in its hellos.) The
"Remaining Time" MUST be set to the current time (in seconds)
before the holding timer on this adjacency is due to expire. If
the corresponding interface is a LAN interface, then the
Restarting Neighbor System ID SHOULD be set to the System ID of
the router from whom the IIH with the RR bit set was received.
This is required to correctly associate the acknowledgement and
holding time in the case where multiple systems on a LAN restart
at approximately the same time. This IIH SHOULD be transmitted
before any LSPs or SNPs are transmitted as a result of the receipt
of the original IIH.
c) if the corresponding interface is a Point-to-Point interface, or
if the receiving router has the highest LnRouterPriority (with
highest source MAC address breaking ties) among those routers to
which the receiving router has an adjacency in state "UP" on this
interface whose IIHs contain the restart TLV, excluding
adjacencies to all routers which are considered in "Restart mode"
(note the actual DIS is NOT changed by this process), initiate the
transmission over the corresponding interface of a complete set of
CSNPs, and set SRMflags on the corresponding interface for all
LSPs in the local LSP database.
Otherwise (i.e., if there was no adjacency in the "UP" state to the
system ID in question), process the IIH as normal by reinitializing
the adjacency and setting the RA bit in the returned IIH.
3.2.2. Use of the SA Bit
The SA bit is used by a starting router to request that its neighbor
suppress advertisement of the adjacency to the starting router in the
neighbor’s LSPs.
A router which is starting has no maintained forwarding function
state. This may or may not be the first time the router has started.
If this is not the first time the router has started, copies of LSPs
generated by this router in its previous incarnation may exist in the
LSP databases of other routers in the network. These copies are
likely to appear "newer" than LSPs initially generated by the
starting router due to the reinitialization of LSP fragment sequence
numbers by the starting router. This may cause temporary blackholes
to occur until the normal operation of the update process causes the
starting router to regenerate and flood copies of its own LSPs with
higher sequence numbers. The temporary blackholes can be avoided if
the starting router’s neighbors suppress advertising an adjacency to
the starting router until the starting router has been able to
propagate newer versions of LSPs generated by previous incarnations.
When a router receives an IIH with the restart TLV having the SA bit
set, if there exists on this interface an adjacency in state "UP"
with the same System ID, and in the case of a LAN circuit, with the
same source LAN address, then the router MUST suppress advertisement
of the adjacency to the neighbor in its own LSPs. Until an IIH with
the SA bit clear has been received, the neighbor advertisement MUST
continue to be suppressed. If the adjacency transitions to the "UP"
state, the new adjacency MUST NOT be advertised until an IIH with the
SA bit clear has been received.
Note that a router which suppresses advertisement of an adjacency
MUST NOT use this adjacency when performing its SPF calculation. In
particular, if an implementation follows the example guidelines
presented in [2] Annex C.2.5 Step 0:b) "pre-load TENT with the local
adjacency database", the suppressed adjacency MUST NOT be loaded into
TENT.
3.3. Adjacency (Re)Acquisition
Adjacency (re)acquisition is the first step in (re)initialization.
Restarting and starting routers will make use of the RR bit in the
restart TLV, though each will use it at different stages of the
(re)start procedure.
3.3.1. Adjacency Reacquisition During Restart
The restarting router explicitly notifies its neighbor that the
adjacency is being reacquired, and hence that it SHOULD NOT
reinitialize the adjacency. This is achieved by setting the RR bit
in the restart TLV. When the neighbor of a restarting router
receives an IIH with the restart TLV having the RR bit set, if there
exists on this interface an adjacency in state "UP" with the same
System ID, and in the case of a LAN circuit, with the same source LAN
address, then the procedures described in 3.2.1 are followed.
A router that does not support the restart capability will ignore the
restart TLV and reinitialize the adjacency as normal, returning an
IIH without the restart TLV.
On restarting, a router initializes the timer T3, starts the timer T2
for each LSPDB, and for each interface (and in the case of a LAN
circuit, for each level) starts the timer T1 and transmits an IIH
containing the restart TLV with the RR bit set.
On a Point-to-Point circuit the restarting router SHOULD set the
"Adjacency Three-Way State" to "Init", because the receipt of the
acknowledging IIH (with RA set) MUST cause the adjacency to enter the
"UP" state immediately.
On a LAN circuit the LAN-ID assigned to the circuit SHOULD be the
same as that used prior to the restart. In particular, for any
circuits for which the restarting router was previously DIS, the use
of a different LAN-ID would necessitate the generation of a new set
of pseudonode LSPs, and corresponding changes in all the LSPs
referencing them from other routers on the LAN. By preserving the
LAN-ID across the restart, this churn can be prevented. To enable a
restarting router to learn the LAN-ID used prior to restart, the
LAN-ID specified in an IIH with RR set MUST be ignored.
Transmission of "normal" IIHs is inhibited until the conditions
described below are met (in order to avoid causing an unnecessary
adjacency initialization). Upon expiry of the timer T1, it is
restarted and the IIH is retransmitted as above.
When a restarting router receives an IIH a local adjacency is
established as usual, and if the IIH contains a restart TLV with the
RA bit set (and on LAN circuits with a Restart Neighbor System ID
which matches that of the local system), the receipt of the
acknowledgement over that interface is noted. When the RA bit is set
and the state of the remote adjacency is "UP", then the timer T3 is
set to the minimum of its current value and the value of the
"Remaining Time" field in the received IIH.
On a Point-to-Point link, receipt of an IIH not containing the
restart TLV is also treated as an acknowledgement, since it indicates
that the neighbor is not restart capable. However, since no CSNP is
guaranteed to be received over this interface, the timer T1 is
cancelled immediately without waiting for a complete set of CSNP(s).
Synchronization may therefore be deemed complete even though there
are some LSPs which are held (only) by this neighbor (see section
3.4). In this case we also want to be certain that the neighbor will
reinitialize the adjacency in order to guarantee that the SRMflags
have been set on its database, thus ensuring eventual LSPDB
synchronization. This is guaranteed to happen except in the case
where the Adjacency Three-Way State in the received IIH is "UP" and
the Neighbor Extended Local Circuit ID matches the extended local
circuit ID assigned by the restarting router. In this case the
restarting router MUST force the adjacency to reinitialize by setting
the local Adjacency Three-Way State to "DOWN" and sending a normal
IIH.
In the case of a LAN interface, receipt of an IIH not containing the
restart TLV is unremarkable since synchronization can still occur so
long as at least one of the non-restarting neighboring routers on the
LAN supports restart. Therefore T1 continues to run in this case.
If none of the neighbors on the LAN are restart capable, T1 will
eventually expire after the locally defined number of retries.
In the case of a Point-to-Point circuit, the "LocalCircuitID" and
"Extended Local Circuit ID" information contained in the IIH can be
used immediately to generate an IIH containing the correct 3-way
handshake information. The presence of "Neighbor Extended Local
Circuit ID" information which does not match the value currently in
use by the local system is ignored (since the IIH may have been
transmitted before the neighbor had received the new value from the
restarting router), but the adjacency remains in the initializing
state until the correct information is received.
In the case of a LAN circuit, the source neighbor information (e.g.,
SNPAAddress) is recorded and used for adjacency establishment and
maintenance as normal.
When BOTH a complete set of CSNP(s) (for each active level, in the
case of a point-to-point circuit) and an acknowledgement have been
received over the interface, the timer T1 is cancelled.
Once the timer T1 has been cancelled, subsequent IIHs are transmitted
according to the normal algorithms, but including the restart TLV
with both RR and RA clear.
If a LAN contains a mixture of systems, only some of which support
the new algorithm, database synchronization is still guaranteed, but
the "old" systems will have reinitialized their adjacencies.
If an interface is active, but does not have any neighboring router
reachable over that interface, the timer T1 would never be cancelled,
and according to clause 3.4.1.1, the SPF would never be run.
Therefore timer T1 is cancelled after some pre-determined number of
expirations (which MAY be 1).
3.3.2. Adjacency Acquisition During Start
The starting router wants to ensure that in the event that a
neighboring router has an adjacency to the starting router in the
"UP" state (from a previous incarnation of the starting router), this
adjacency is reinitialized. The starting router also wants
neighboring routers to suppress advertisement of an adjacency to the
starting router until LSP database synchronization is achieved. This
is achieved by sending IIHs with the RR bit clear and the SA bit set
in the restart TLV. The RR bit remains clear and the SA bit remains
set in subsequent transmissions of IIHs until the adjacency has
reached the "UP" state and the initial T1 timer interval (see below)
has expired.
Receipt of an IIH with the RR bit clear will result in the
neighboring router utilizing normal operation of the adjacency state
machine. This will ensure that any old adjacency on the neighboring
router will be reinitialized.
Upon receipt of an IIH with the SA bit set, the behavior described in
3.2.2 is followed.
Upon starting, a router starts timer T2 for each LSPDB.
For each interface (and in the case of a LAN circuit, for each
level), when an adjacency reaches the "UP" state, the starting router
starts a timer T1 and transmits an IIH containing the restart TLV
with the RR bit clear and SA bit set. Upon expiry of the timer T1,