↓ Skip to main content
KAZULOG.INFO
  1. Network Articles/
  2. MPLS Articles/

LDP-IGP Synchronization and LDP Session Protection

Table of Contents

LDP-IGP Synchronization and LDP Session Protection

The IGP that distributes routes and LDP that distributes labels run independently. When the IGP picks a link on which LDP is not usable, there is no label and only the MPLS traffic is dropped. This article explains that gap and the two mechanisms that close it, then measures both in an IOS XR (XRd) lab. The basics of LDP are covered in What Is LDP.

What happens when a link without labels is chosen

MPLS forwarding assumes that LDP has finished distributing the label for the FEC towards the next hop the IGP chose. Two situations break that assumption.

The first is LDP not running on the link at all. A missing configuration, an ACL or filter, or a mismatched password stops the session from coming up; the IGP keeps choosing the link and only the MPLS traffic is dropped, permanently.

The second is the convergence gap right after a link comes up, but which of the two finishes first depends on the timers. With the IOS XR defaults, LDP sends Hellos every 5 seconds and the session is established over TCP immediately, while OSPF sends Hellos every 10 seconds and reaches FULL through ExStart / Exchange / Loading. On a point-to-point link LDP usually comes up first (measured later in this article), and in that case restoring a link does not create a blackhole.

How things break also depends on the kind of traffic. Plain IP is sent out without a label, and if the next router has an IP route it is forwarded anyway. VPN (VPNv4) traffic, on the other hand, requires the transport label and has no way to fall back to IP forwarding. The same fault drops only the VPN traffic.

LDP-IGP synchronization

RFC 5443 defines a mechanism where the IGP advertises a link at maximum cost until LDP is “fully operational” on it. Because the cost is the maximum, any other path wins and traffic does not flow onto a link that has no labels.

Fully operational means all three conditions hold (section 2).

ConditionContent
1An LDP hello adjacency exists on the link
2A session matching the hello adjacency’s LDP Identifier is established with the peer
3All label bindings have been exchanged over that session

A router cannot verify the third condition. Section 2 of RFC 5443 says it “cannot generally be verified by a router and some estimation may have to be used”, and points to a configurable hold-down timer as a simple implementation strategy. The mechanism that makes it verifiable is End-of-LIB in RFC 5919, which signals that label advertisement is complete; with it the timer is unnecessary. On IOS XR the timer is igp sync delay on-session-up <seconds> and shows up as Sync delay in show mpls ldp igp sync (measured later in this article).

The maximum cost differs per IGP.

IGPValueNote
OSPF65535 (LSInfinity, 16 bits)—
IS-IS16777214 (2^24−2)2^24−1 would remove the link from the topology, so it is avoided

There are two caveats. It is meaningless without an alternate path — if there is no other route, the link is chosen anyway. Section 3 of RFC 5443 likewise says the mechanism is for cases where “availability of LDP-signaled LSPs matters more than always choosing the optimal path”. Also, on a broadcast link the whole link is costed out, not the individual peer (section 3). When TE tunnels are in use, the mechanism applies only to the IP link cost and not to the TE metric (section 4), since disturbing the TE path computation would cause unnecessary tunnel reroutes.

LDP session protection

An LDP session is TCP between Lo0s, but whether it may be kept depends on whether a hello adjacency is alive. When the link fails and every hello adjacency disappears, the session goes down too and the label bindings are lost. On restoration it all starts again: Hello → TCP → Initialization → label exchange.

LDP session protection fills that gap with targeted Hellos. They are the Extended Discovery messages of RFC 5036 section 2.4.2 and, unlike link Hellos (multicast), are sent as unicast to the peer’s address. Even with the link gone, the hello adjacency survives as long as an IP path to the peer remains, and so do the session and its labels. Extended Discovery is asymmetric: one side starts sending and the receiver decides whether to answer.

What session protection changes is not the speed of recovery but whether the session and its labels are lost. It does not help if the peer cannot be reached by IP.

Lab verification

Lab setup

The core is doubled up, and PE1 and P1 are joined by a single link. Stopping discovery on that one link prevents an LDP session between PE1 and P1.

LinkSubnetOSPF costRole
PE1 - P110.2.3.0/241Under test; the ACL goes here
PE1 - P210.2.4.0/24100Entry to the detour
P1 - P3 / P2 - P410.3.5.0/24 / 10.4.6.0/241The two planes
P3 - PE210.5.7.0/241—
P4 - PE210.6.7.0/24100Exit from the detour
P1 - P2 / P3 - P410.3.4.0/24 / 10.5.6.0/241Cross links; the detour for session protection

By cost, PE1 to PE2 is 3 via PE1-P1-P3-PE2, 103 via PE1-P2-P1-P3-PE2, and 201 via PE1-P2-P4-PE2.

CE1 and CE2 attach to vrf CUST-A (RD/RT 65001:1) on the PEs over eBGP, and PE1 and PE2 run iBGP vpnv4 between their Loopback0 addresses. Customer traffic always carries two labels, so it has no IP-forwarding escape route.

How the outage is measured

The forwarding outage is measured with a ping from CE1 that runs across the configuration change (count 600 interval 100 timeout 1). In the default collection order (clear the counters, ping, then show), the ping runs after the change and the convergence wait, so it only sees the steady state. The ..._flap.txt of each STEP holds that record.

The loss count only resolves times coarser than the ping’s own granularity. With timeout 1, one lost packet means the next is sent a second later, so an outage of 150 ms and one of 1 s both show up as “one packet lost”. This lab treats “how many packets were lost” and “which link carried the traffic” as the facts, and does not claim a millisecond switchover time.

XRd forwards in software on CML. Read the numbers as relative comparisons within this lab.

Verification steps

STEP 0 to 3 cover LDP-IGP synchronization, and STEP 9 to 12 the measurements across a link flap (STEP 4 to 8 are a first pass of the same operations; their files are kept).

STEPActionWhat to confirm
0Initial stateTraffic goes PE1-P1-P3-PE2. The LDP capabilities do not include End-of-LIB
1ACL (deny UDP 646) on both ends of PE1 - P1No session forms and the VPN goes down, while plain IP still works
2mpls ldp sync on both endsThe link is advertised with cost 65535 and traffic recovers over the detour
3Remove the ACLSynchronization completes and the cost returns
9Flap the link with no sync and no protectionThe baseline outage; which of LDP and OSPF comes up first
10Same with syncThe difference from STEP 9
11Same with protection addedThe session survives and targeted hellos cross the detour
12Same with protection off and the hold-down timer onThe cost stays at 65535 until synchronization is declared

When LDP is down, only the VPN breaks

Both ends of PE1 - P1 get an ACL that drops only LDP discovery. TCP 646 is left open, because without discovery no session forms anyway.

Configuration committed on PE1 in STEP 1
RP/0/RP0/CPU0:PE1#show configuration commit changes last 1
Sun Sep 27 21:52:06.954 UTC
!! Building configuration...
!! IOS XR Configuration 26.1.1
ipv4 access-list BLOCK-LDP-DISCOVERY
 10 deny udp any any eq ldp
 20 permit ipv4 any any
!
interface GigabitEthernet0/0/0/1
 ipv4 access-group BLOCK-LDP-DISCOVERY ingress
!
end

After the 15-second hold time the session with P1 disappears, leaving only the one with P2.

STEP 1 PE1 show mpls ldp neighbor brief
RP/0/RP0/CPU0:PE1#show mpls ldp neighbor brief
Sun Sep 27 21:57:48.699 UTC

Peer               GR  NSR  Up Time     Discovery   Addresses     Labels
                                        ipv4  ipv6  ipv4  ipv6  ipv4   ipv6
-----------------  --  ---  ----------  ----------  ----------  ------------
4.4.4.4:0          N   N    00:17:26    1     0     4     0     14     0

All customer traffic is gone. The traceroute does not advance a single hop past PE1: !N is network unreachable, returned by PE1 itself.

STEP 1 CE1 traceroute (to 192.168.2.1)
RP/0/RP0/CPU0:CE1#traceroute 192.168.2.1 source 192.168.1.1 timeout 1 probe 2 maxttl 8
Sun Sep 27 21:55:46.047 UTC

Type escape sequence to abort.
Tracing the route to 192.168.2.1

 1  10.1.2.2 !N  *

The forwarding entry gives the reason. The VPN label is known, but no transport label is available and the next hop became drop.

STEP 1 PE1 show cef vrf CUST-A 192.168.2.0/24 detail (tail)
RP/0/RP0/CPU0:PE1#show cef vrf CUST-A 192.168.2.0/24 detail
Sun Sep 27 21:57:58.538 UTC
192.168.2.0/24, version 7, internal 0x5000001 0x30 (ptr 0x885192c0) [1], 0x0 (0x0), 0x208 (0x899cb518)
 Updated Sep 27 21:41:43.111
 Prefix Len 24, traffic index 0, precedence n/a, priority 3
  gateway array (0x8938ba58) reference count 1, flags 0x203a, source rib (7), 0 backups
                [1 type 1 flags 0xc0441 (0x89a044b8) ext 0x0 (0x0)]
  LW-LDI[type=0, refc=0, ptr=0x0, sh-ldi=0x0]
  gateway array update type-time 3 Sep 27 21:57:51.444
 LDI Update time Sep 27 21:52:21.357
   via 7.7.7.7/32, 0 dependencies, recursive [flags 0x6000]
    path-idx 0 NHID 0x0 [0x885295e0 0x0]
    recursion-via-/32
    next hop VRF - 'default', table - 0xe0000000
    unresolved
     labels imposed {24011}

    Load distribution: 0 (refcount 1)

    Hash  OK  Interface                 Address
    0     Y   recursive                 drop

Plain IP, however, still gets through. Every ping from PE1 to PE2’s Loopback0 succeeds, and the first hop carries no label: PE1 sends it as IP and P1, which has an IP route, forwards it. Whether the traffic can fall back to IP forwarding is what separates the VPN from plain IP.

STEP 1 PE1 ping / traceroute (to 7.7.7.7)
RP/0/RP0/CPU0:PE1#ping 7.7.7.7 source 2.2.2.2 count 50 timeout 1
Sun Sep 27 21:55:50.945 UTC
Type escape sequence to abort.
Sending 50, 100-byte ICMP Echos to 7.7.7.7 timeout is 1 seconds:
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Success rate is 100 percent (50/50), round-trip min/avg/max = 10/12/47 ms
RP/0/RP0/CPU0:PE1#traceroute 7.7.7.7 source 2.2.2.2 timeout 1 probe 2 maxttl 6
Sun Sep 27 21:55:52.371 UTC

Type escape sequence to abort.
Tracing the route to 7.7.7.7

 1  10.2.3.3 7 msec  5 msec
 2  10.3.5.5 [MPLS: Label 24008 Exp 0] 11 msec  11 msec
 3  10.5.7.7 13 msec  *

Advertising the link with the maximum cost

mpls ldp sync goes on that interface in OSPF on PE1 and P1.

Configuration committed on PE1 in STEP 2
RP/0/RP0/CPU0:PE1#show configuration commit changes last 1
Sun Sep 27 22:02:02.263 UTC
!! Building configuration...
!! IOS XR Configuration 26.1.1
router ospf 1
 area 0
  interface GigabitEthernet0/0/0/1
   mpls ldp sync
  !
 !
!
end

The synchronization state is shown per interface, including which condition is unmet. No hello adjacency means condition 1 of fully operational is missing. Gi0/0/0/2, which has no ACL, reads Ready in the same output.

STEP 2 PE1 show mpls ldp igp sync
RP/0/RP0/CPU0:PE1#show mpls ldp igp sync
Sun Sep 27 22:05:52.251 UTC

GigabitEthernet0/0/0/1:
  VRF: 'default' (0x60000000)
  Sync delay: Disabled
  Sync status: Not ready (No hello adjacency)

GigabitEthernet0/0/0/2:
  VRF: 'default' (0x60000000)
  Sync delay: Disabled
  Sync status: Ready
    Peers:
      4.4.4.4:0

The advertised value appears in the Router-LSA: only the link to P1 (3.3.3.3) is 65535.

STEP 2 PE1 show ospf database router self-originate (excerpt)
RP/0/RP0/CPU0:PE1#show ospf database router self-originate
Sun Sep 27 22:05:47.791 UTC


            OSPF Router with ID (2.2.2.2) (Process ID 1)

		Router Link States (Area 0)

  LS age: 226
  Options: (No TOS-capability, DC)
  LS Type: Router Links
  Link State ID: 2.2.2.2
  Advertising Router: 2.2.2.2
  LS Seq Number: 80000005
  Checksum: 0xe3cd
  Length: 84
   Number of Links: 5

    Link connected to: a Stub Network
     (Link ID) Network/subnet number: 2.2.2.2
     (Link Data) Network Mask: 255.255.255.255
      Number of TOS metrics: 0
       TOS 0 Metrics: 1

    Link connected to: another Router (point-to-point)
     (Link ID) Neighboring Router ID: 3.3.3.3
     (Link Data) Router Interface address: 10.2.3.2
      Number of TOS metrics: 0
       TOS 0 Metrics: 65535

    Link connected to: a Stub Network
     (Link ID) Network/subnet number: 10.2.3.0
     (Link Data) Network Mask: 255.255.255.0
      Number of TOS metrics: 0
       TOS 0 Metrics: 1

    Link connected to: another Router (point-to-point)
     (Link ID) Neighboring Router ID: 4.4.4.4
     (Link Data) Router Interface address: 10.2.4.2
      Number of TOS metrics: 0
       TOS 0 Metrics: 100

    Link connected to: a Stub Network
     (Link ID) Network/subnet number: 10.2.4.0
     (Link Data) Network Mask: 255.255.255.0
      Number of TOS metrics: 0
       TOS 0 Metrics: 100

Traffic moves to the cost-100 link and detours while keeping both labels: out to P2 (10.2.4.4), back to P1 (10.3.4.3) over the cross link, and on along the original path.

STEP 2 CE1 traceroute
RP/0/RP0/CPU0:CE1#traceroute 192.168.2.1 source 192.168.1.1 timeout 1 probe 2 maxttl 8
Sun Sep 27 22:04:49.702 UTC

Type escape sequence to abort.
Tracing the route to 192.168.2.1

 1  10.1.2.2 6 msec  5 msec
 2  10.2.4.4 [MPLS: Labels 24009/24011 Exp 0] 22 msec  21 msec
 3  10.3.4.3 [MPLS: Labels 24009/24011 Exp 0] 21 msec  23 msec
 4  10.3.5.5 [MPLS: Labels 24008/24011 Exp 0] 22 msec  19 msec
 5  10.5.7.7 [MPLS: Label 24011 Exp 0] 20 msec  19 msec
 6  10.7.8.8 19 msec  *

Sync delay: Disabled corresponds to the third condition of fully operational (all label bindings exchanged) being only an estimate. Section 2 of RFC 5443 states that “at the present time, the latter condition cannot generally be verified by a router and some estimation may have to be used”, and that “a simple implementation strategy is to use a configurable hold-down timer”. The mechanism that makes it verifiable is End-of-LIB in RFC 5919, which signals that label advertisement is complete; once implemented, the timer is no longer needed.

The XRd in this lab does not advertise End-of-LIB. It is absent from the LDP capabilities, so the estimating path applies.

STEP 0 PE1 show mpls ldp capabilities
RP/0/RP0/CPU0:PE1#show mpls ldp capabilities
Sun Sep 27 21:47:53.665 UTC
  Type     Description                                Owner
  ------   ----------------------------------------   ------------
  0x50b    Typed Wildcard FEC                         LDP
  0x3eff   Cisco IOS-XR                               LDP
  0x508    MP: Point-to-Multipoint (P2MP)             mLDP
  0x509    MP: Multipoint-to-Multipoint (MP2MP)       mLDP
  0x50c    MP: Multi-Topology (MT)                    mLDP
  0x703    P2MP PW                                    L2VPN-AToM

Removing the ACL in STEP 3 completes the synchronization. The trace records the moment it is declared, and the cost returns to 1.

STEP 3 PE1 show mpls ldp trace igp-sync (tail)
RP/0/RP0/CPU0:PE1#show mpls ldp trace igp-sync
Sun Sep 27 22:14:09.547 UTC
5 wrapping entries (1027456 possible, 3200 allocated, 826 filtered, 831 total)
Sep 27 21:40:38.884 mpls/ldp/isyn 0/RP0/CPU0 t4113  [ISYNC]:2158: Intf GigabitEthernet0/0/0/1 (ifh 0x18): Nbr 3.3.3.3:0 came up, sync_achieved up
Sep 27 21:40:43.190 mpls/ldp/isyn 0/RP0/CPU0 t4113  [ISYNC]:2158: Intf GigabitEthernet0/0/0/2 (ifh 0x20): Nbr 4.4.4.4:0 came up, sync_achieved up
Sep 27 21:52:20.343 mpls/ldp/isyn 0/RP0/CPU0 t4113  [ISYNC]:1472: Intf GigabitEthernet0/0/0/1 (ifh 0x18): Adj 10.2.3.3 is being deleted, sync achieved 0
Sep 27 21:52:20.343 mpls/ldp/isyn 0/RP0/CPU0 t4113  [ISYNC]:1476: Adj 10.2.3.3 being deleted belongs to adj_grp 3.3.3.3, has adj_flags 0xb4
Sep 27 22:10:09.824 mpls/ldp/isyn 0/RP0/CPU0 t4113  [ISYNC]:2158: Intf GigabitEthernet0/0/0/1 (ifh 0x18): Nbr 3.3.3.3:0 came up, sync_achieved up
STEP 3 PE1 show ospf database router self-originate (excerpt)
RP/0/RP0/CPU0:PE1#show ospf database router self-originate
Sun Sep 27 22:13:44.803 UTC


            OSPF Router with ID (2.2.2.2) (Process ID 1)

		Router Link States (Area 0)

  LS age: 215
  Options: (No TOS-capability, DC)
  LS Type: Router Links
  Link State ID: 2.2.2.2
  Advertising Router: 2.2.2.2
  LS Seq Number: 80000006
  Checksum: 0xffaf
  Length: 84
   Number of Links: 5

    Link connected to: a Stub Network
     (Link ID) Network/subnet number: 2.2.2.2
     (Link Data) Network Mask: 255.255.255.255
      Number of TOS metrics: 0
       TOS 0 Metrics: 1

    Link connected to: another Router (point-to-point)
     (Link ID) Neighboring Router ID: 3.3.3.3
     (Link Data) Router Interface address: 10.2.3.2
      Number of TOS metrics: 0
       TOS 0 Metrics: 1

    Link connected to: a Stub Network
     (Link ID) Network/subnet number: 10.2.3.0
     (Link Data) Network Mask: 255.255.255.0
      Number of TOS metrics: 0
       TOS 0 Metrics: 1

    Link connected to: another Router (point-to-point)
     (Link ID) Neighboring Router ID: 4.4.4.4
     (Link Data) Router Interface address: 10.2.4.2
      Number of TOS metrics: 0
       TOS 0 Metrics: 100

    Link connected to: a Stub Network
     (Link ID) Network/subnet number: 10.2.4.0
     (Link Data) Network Mask: 255.255.255.0
      Number of TOS metrics: 0
       TOS 0 Metrics: 100

On a link recovery, synchronization makes no difference

Both ends of PE1 - P1 were shut down and restored 40 seconds later, once without synchronization (STEP 9) and once with it (STEP 10). Both lost exactly one packet; there is no difference.

The syslog timestamps give the reason: LDP comes up before OSPF.

STEP 9 PE1 show logging (the two recovery lines)
RP/0/RP0/CPU0:Sep 27 19:14:28.310 UTC: mpls_ldp[1178]: %ROUTING-LDP-5-NBR_CHANGE : VRF 'default' (0x60000000), Neighbor 3.3.3.3:0 is UP (IPv4 connection)
RP/0/RP0/CPU0:Sep 27 19:14:30.424 UTC: ospf[1035]: %ROUTING-OSPF-5-ADJCHG : Process 1, Nbr 3.3.3.3 on GigabitEthernet0/0/0/1 in area 0 from LOADING to FULL, Loading Done, vrf default vrfid 0x60000000

By the time the IGP picks the link again, labels are already available, so there is no moment for synchronization to hold back. With IOS XR defaults, the LDP session is established over TCP immediately, while OSPF waits for its 10-second Hello and then goes through ExStart, Exchange and Loading.

Synchronization is not a feature that speeds up recovery; it is insurance against the IGP choosing a link where LDP cannot run. It earns its place when LDP fails to come up because of a missing configuration, a filter or a password mismatch (STEP 1 to 2 is exactly that), or when LDP converges later than the IGP because of a large label table or a slow peer.

Session protection: keeping the session when the link is gone

With session protection (STEP 11), the same operation never drops the LDP session. Not one NBR_CHANGE appears, and the label bindings stay.

show mpls ldp discovery shows why: next to the link hello adjacency there is a Targeted Hellos entry, and its hold time is 90 seconds rather than 15.

STEP 11 PE1 show mpls ldp discovery
RP/0/RP0/CPU0:PE1#show mpls ldp discovery
Sun Sep 27 19:37:43.015 UTC

Local LDP Identifier: 2.2.2.2:0
Discovery Sources:
  Interfaces:
    GigabitEthernet0/0/0/1 : xmit/recv
      VRF: 'default' (0x60000000)
      LDP Id: 3.3.3.3:0, Transport address: 3.3.3.3
          Hold time: 15 sec (local:15 sec, peer:15 sec)
          Established: Sep 27 19:33:39.109 (00:04:03 ago)

    GigabitEthernet0/0/0/2 : xmit/recv
      VRF: 'default' (0x60000000)
      LDP Id: 4.4.4.4:0, Transport address: 4.4.4.4
          Hold time: 15 sec (local:15 sec, peer:15 sec)
          Established: Sep 27 19:07:06.757 (00:30:36 ago)

  Targeted Hellos:
    2.2.2.2 -> 3.3.3.3 (active), xmit/recv
      LDP Id: 3.3.3.3:0
          Hold time: 90 sec (local:90 sec, peer:90 sec)
          Established: Sep 27 19:32:25.537 (00:05:17 ago)

    2.2.2.2 -> 4.4.4.4 (active), xmit
          Established: N/A

The capture shows where they went: during the outage, unicast hellos crossed the detour (PE1 - P2), addressed to the peer’s Loopback0 (3.3.3.3).

STEP 11 No.204 targeted hello (PE1 → P1, over the detour) tshark -V excerpt
Internet Protocol Version 4, Src: 2.2.2.2, Dst: 3.3.3.3
    0100 .... = Version: 4
    .... 0101 = Header Length: 20 bytes (5)
    Differentiated Services Field: 0xc0 (DSCP: CS6, ECN: Not-ECT)
        1100 00.. = Differentiated Services Codepoint: Class Selector 6 (48)
        .... ..00 = Explicit Congestion Notification: Not ECN-Capable Transport (0)
    Total Length: 70
    Identification: 0xb900 (47360)
    000. .... = Flags: 0x0
        0... .... = Reserved bit: Not set
        .0.. .... = Don't fragment: Not set
        ..0. .... = More fragments: Not set
    ...0 0000 0000 0000 = Fragment Offset: 0
    Time to Live: 255
    Protocol: UDP (17)
    Header Checksum: 0xf7dc [validation disabled]
    [Header checksum status: Unverified]
    Source Address: 2.2.2.2
    Destination Address: 3.3.3.3
    [Stream index: 5]
User Datagram Protocol, Src Port: 646, Dst Port: 646
    Source Port: 646
    Destination Port: 646
    Length: 50
    Checksum: 0x5cc5 [unverified]
    [Checksum Status: Unverified]
    [Stream index: 3]
    [Stream Packet Number: 2]
    [Timestamps]
        [Time since first frame: 253.577000 milliseconds]
        [Time since previous frame: 253.577000 milliseconds]
    UDP payload (42 bytes)
Label Distribution Protocol
    Version: 1
    PDU Length: 38
    LSR ID: 2.2.2.2
    Label Space ID: 0
    Hello Message
        0... .... = U bit: Unknown bit not set
        Message Type: Hello Message (0x100)
        Message Length: 28
        Message ID: 0x00000000
        Common Hello Parameters
            00.. .... = TLV Unknown bits: Known TLV, do not Forward (0x0)
            TLV Type: Common Hello Parameters (0x400)
            TLV Length: 4
            Hold Time: 90
            1... .... .... .... = Targeted Hello: Targeted Hello
            .1.. .... .... .... = Hello Requested: Source requests periodic hellos
Download the pcap of the packet in the tshark output above (No.204 targeted hello)

The Hello Requested bit is the asymmetric exchange defined in section 2.4.2 of RFC 5036: the initiating LSR asks the other side to send periodic hellos back.

What protection changes is whether the session and the labels are lost, not the length of the outage (still one packet). In this topology the traffic moves to the detour as soon as the link goes down, so the cost of the switch is the same under every condition. Keeping the labels avoids re-establishing the session and re-exchanging them.

The hold-down timer: waiting before declaring synchronization

On IOS XR, igp sync delay on-session-up <seconds> under mpls ldp sets how long to wait after the session comes up before declaring synchronization. STEP 12 sets 30 seconds and flaps the link with protection off, so that the session really drops.

Looking at PE1 right after the recovery, while synchronization is Deferred, the Router-LSA still advertises the link to P1 with 65535.

STEP 12 PE1 right after the recovery (+5 s and +15 s)
Sun Sep 27 19:43:16.685 UTC
RP/0/RP0/CPU0:PE1#show clock

Sun Sep 27 19:43:16.963 UTC
19:43:17.006 UTC Sun Sep 27 2026
RP/0/RP0/CPU0:PE1#show mpls ldp igp sync

Sun Sep 27 19:43:17.081 UTC

GigabitEthernet0/0/0/1:
  VRF: 'default' (0x60000000)
  Sync delay: 30 sec
  Sync status: Deferred   (20 sec remaining)

GigabitEthernet0/0/0/2:
  VRF: 'default' (0x60000000)
  Sync delay: 30 sec
  Sync status: Ready
    Peers:
      4.4.4.4:0


RP/0/RP0/CPU0:PE1#show ospf database router self-originate

Sun Sep 27 19:43:17.202 UTC


            OSPF Router with ID (2.2.2.2) (Process ID 1)

		Router Link States (Area 0)

  LS age: 8
  Options: (No TOS-capability, DC)
  LS Type: Router Links
  Link State ID: 2.2.2.2
  Advertising Router: 2.2.2.2
  LS Seq Number: 80000010
  Checksum: 0xcdd8
  Length: 84
   Number of Links: 5

    Link connected to: a Stub Network
     (Link ID) Network/subnet number: 2.2.2.2
     (Link Data) Network Mask: 255.255.255.255
      Number of TOS metrics: 0
       TOS 0 Metrics: 1

    Link connected to: another Router (point-to-point)
     (Link ID) Neighboring Router ID: 3.3.3.3
     (Link Data) Router Interface address: 10.2.3.2
      Number of TOS metrics: 0
       TOS 0 Metrics: 65535

    Link connected to: a Stub Network
     (Link ID) Network/subnet number: 10.2.3.0
     (Link Data) Network Mask: 255.255.255.0
      Number of TOS metrics: 0
       TOS 0 Metrics: 1

    Link connected to: another Router (point-to-point)
     (Link ID) Neighboring Router ID: 4.4.4.4
     (Link Data) Router Interface address: 10.2.4.2
      Number of TOS metrics: 0
       TOS 0 Metrics: 100

    Link connected to: a Stub Network
     (Link ID) Network/subnet number: 10.2.4.0
     (Link Data) Network Mask: 255.255.255.0
      Number of TOS metrics: 0
       TOS 0 Metrics: 100


RP/0/RP0/CPU0:PE1#show route 7.7.7.7/32

Sun Sep 27 19:43:18.000 UTC

Routing entry for 7.7.7.7/32
  Known via "ospf 1", distance 110, metric 104, type intra area
  Installed Sep 27 19:42:07.212 for 00:01:10
  Routing Descriptor Blocks
    10.2.4.4, from 7.7.7.7, via GigabitEthernet0/0/0/2
      Route metric is 104
  No advertising protos.

The later snapshot is back to Ready with a cost of 1. This is the part RFC 5443 calls an estimate, filled in by the implementation with a timer.

The four conditions side by side

STEPSyncProtectionDelayPing across the changeLDP sessionTargeted hellos on the detour
9NoNoNo599/600 (1 lost)Dropped0
10YesNoNo599/600 (1 lost)Dropped0
11YesYesNo599/600 (1 lost)Kept6
12YesNoYes599/600 (1 lost)Dropped0
Synchronization only made a difference in STEP 1 to 2, where reachability went from 0% to 100%. Across a link flap (STEP 9 to 12), one packet is lost whichever of synchronization, protection and the delay is enabled. What changes is whether the session and labels are lost (protection) and how long the declaration is held back (the delay).
CE1 to CE2 (STEP 0, 50 packets after convergence)
RP/0/RP0/CPU0:CE1#ping 192.168.2.1 source 192.168.1.1 count 50 timeout 1
Sun Sep 27 21:46:49.592 UTC
Type escape sequence to abort.
Sending 50, 100-byte ICMP Echos to 192.168.2.1 timeout is 1 seconds:
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Success rate is 100 percent (50/50), round-trip min/avg/max = 16/22/81 ms

Verification config and show output

For every STEP the following files were collected from all eight routers, one file per router. The verification config is the ..._run.txt (the final state is the one from STEP 13). Each STEP was collected in the order “clear the counters, ping, then show”.

FileContents
..._clear.txtThe counters cleared at the start of the STEP
..._ping.txtping (50 packets) and traceroute after convergence: between the CEs for customer traffic, between Loopback0 for the core
..._flap.txtThe ping that ran across the configuration change (600 packets at 100 ms), from CE1, on the link-flap STEPs only
..._show.txtshow version / show route / the OSPF set / the LDP set (igp sync / discovery / neighbor / bindings / capabilities) / MPLS forwarding / VPN and BGP
..._log.txtshow logging limited to the range of that STEP
..._run.txtshow running-config at that STEP (that is, the verification config for the STEP)
..._trace.txtThe six core routers only: show mpls ldp trace igp-sync / peer / binding and show mpls lsd trace
..._commit.cfgThe configuration committed in that STEP. On the flap STEPs it is split between taking the link down (pe1 / p1) and bringing it back (pe1r / p1r)
..._during.txtFlap STEPs only: PE1 right after the recovery (+5 s / +15 s), its synchronization state and Router-LSA

STEP 0: Initial state

Routershowpingoutageclearsyslogrunning-configtracecommitted configjust after recovery
CE1showping—clearlogrun———
PE1showping—clearlogruntrace——
P1showping—clearlogruntrace——
P2showping—clearlogruntrace——
P3showping—clearlogruntrace——
P4showping—clearlogruntrace——
PE2showping—clearlogruntrace——
CE2showping—clearlogrun———

STEP 1: ACL (deny UDP 646) on both ends of PE1 - P1

Routershowpingoutageclearsyslogrunning-configtracecommitted configjust after recovery
CE1showping—clearlogrun———
PE1showping—clearlogruntracepe1—
P1showping—clearlogruntracep1—
P2showping—clearlogruntrace——
P3showping—clearlogruntrace——
P4showping—clearlogruntrace——
PE2showping—clearlogruntrace——
CE2showping—clearlogrun———

STEP 2: mpls ldp sync on both ends

Routershowpingoutageclearsyslogrunning-configtracecommitted configjust after recovery
CE1showping—clearlogrun———
PE1showping—clearlogruntracepe1—
P1showping—clearlogruntracep1—
P2showping—clearlogruntrace——
P3showping—clearlogruntrace——
P4showping—clearlogruntrace——
PE2showping—clearlogruntrace——
CE2showping—clearlogrun———

STEP 3: Remove the ACL

Routershowpingoutageclearsyslogrunning-configtracecommitted configjust after recovery
CE1showping—clearlogrun———
PE1showping—clearlogruntracepe1—
P1showping—clearlogruntracep1—
P2showping—clearlogruntrace——
P3showping—clearlogruntrace——
P4showping—clearlogruntrace——
PE2showping—clearlogruntrace——
CE2showping—clearlogrun———

STEP 4: Remove sync, then flap the link

Routershowpingoutageclearsyslogrunning-configtracecommitted configjust after recovery
CE1showpingoutageclearlogrun———
PE1showping—clearlogruntracepe1 pe1rduring
P1showping—clearlogruntracep1 p1r—
P2showping—clearlogruntrace——
P3showping—clearlogruntrace——
P4showping—clearlogruntrace——
PE2showping—clearlogruntrace——
CE2showping—clearlogrun———

STEP 5: Same with sync

Routershowpingoutageclearsyslogrunning-configtracecommitted configjust after recovery
CE1showpingoutageclearlogrun———
PE1showping—clearlogruntracepe1 pe1rduring
P1showping—clearlogruntracep1 p1r—
P2showping—clearlogruntrace——
P3showping—clearlogruntrace——
P4showping—clearlogruntrace——
PE2showping—clearlogruntrace——
CE2showping—clearlogrun———

STEP 6: Same with session protection added

Routershowpingoutageclearsyslogrunning-configtracecommitted configjust after recovery
CE1showpingoutageclearlogrun———
PE1showping—clearlogruntracepe1 pe1rduring
P1showping—clearlogruntracep1 p1r—
P2showping—clearlogruntrace——
P3showping—clearlogruntrace——
P4showping—clearlogruntrace——
PE2showping—clearlogruntrace——
CE2showping—clearlogrun———

STEP 7: Same with the hold-down timer

Routershowpingoutageclearsyslogrunning-configtracecommitted configjust after recovery
CE1showpingoutageclearlogrun———
PE1showping—clearlogruntracepe1 pe1rduring
P1showping—clearlogruntracep1 p1r—
P2showping—clearlogruntrace——
P3showping—clearlogruntrace——
P4showping—clearlogruntrace——
PE2showping—clearlogruntrace——
CE2showping—clearlogrun———

STEP 8: Remove protection, delay and sync

Routershowpingoutageclearsyslogrunning-configtracecommitted configjust after recovery
CE1showping—clearlogrun———
PE1showping—clearlogruntracepe1—
P1showping—clearlogruntracep1—
P2showping—clearlogruntrace——
P3showping—clearlogruntrace——
P4showping—clearlogruntrace——
PE2showping—clearlogruntrace——
CE2showping—clearlogrun———

STEP 9: Flap the link with no sync and no protection

Routershowpingoutageclearsyslogrunning-configtracecommitted configjust after recovery
CE1showpingoutageclearlogrun———
PE1showping—clearlogruntracepe1 pe1rduring
P1showping—clearlogruntracep1 p1r—
P2showping—clearlogruntrace——
P3showping—clearlogruntrace——
P4showping—clearlogruntrace——
PE2showping—clearlogruntrace——
CE2showping—clearlogrun———

STEP 10: Same with sync

Routershowpingoutageclearsyslogrunning-configtracecommitted configjust after recovery
CE1showpingoutageclearlogrun———
PE1showping—clearlogruntracepe1 pe1rduring
P1showping—clearlogruntracep1 p1r—
P2showping—clearlogruntrace——
P3showping—clearlogruntrace——
P4showping—clearlogruntrace——
PE2showping—clearlogruntrace——
CE2showping—clearlogrun———

STEP 11: Same with sync and protection

Routershowpingoutageclearsyslogrunning-configtracecommitted configjust after recovery
CE1showpingoutageclearlogrun———
PE1showping—clearlogruntracepe1 pe1rduring
P1showping—clearlogruntracep1 p1r—
P2showping—clearlogruntrace——
P3showping—clearlogruntrace——
P4showping—clearlogruntrace——
PE2showping—clearlogruntrace——
CE2showping—clearlogrun———

STEP 12: Same with sync, no protection and the delay

Routershowpingoutageclearsyslogrunning-configtracecommitted configjust after recovery
CE1showpingoutageclearlogrun———
PE1showping—clearlogruntracepe1 pe1rduring
P1showping—clearlogruntracep1 p1r—
P2showping—clearlogruntrace——
P3showping—clearlogruntrace——
P4showping—clearlogruntrace——
PE2showping—clearlogruntrace——
CE2showping—clearlogrun———

STEP 13: Remove sync and delay (final state)

Routershowpingoutageclearsyslogrunning-configtracecommitted configjust after recovery
CE1showping—clearlogrun———
PE1showping—clearlogruntracepe1—
P1showping—clearlogruntracep1—
P2showping—clearlogruntrace——
P3showping—clearlogruntrace——
P4showping—clearlogruntrace——
PE2showping—clearlogruntrace——
CE2showping—clearlogrun———

Packet captures were taken per STEP on the three segments.

STEPCE1 - PE1PE1 - P1PE1 - P2 (detour)
0pcappcappcap
1pcappcappcap
2pcappcappcap
3pcappcappcap
4pcappcappcap
5pcappcappcap
6pcappcappcap
7pcappcappcap
8pcappcappcap
9pcappcappcap
10pcappcappcap
11pcappcappcap
12pcappcappcap
13pcappcappcap

References

RFCTitleSummary
RFC 5443LDP IGP SynchronizationThe IGP advertises the maximum cost until LDP is fully operational. The definition of fully operational and the cost values (section 2); applicability and broadcast links (section 3).
RFC 5036LDP SpecificationBasic Discovery (section 2.4.1) and Extended Discovery (section 2.4.2), which defines the asymmetric targeted Hello exchange.

Book: Luc De Ghein, MPLS Fundamentals (Cisco Press, 2006), Chapter 4

Related articles