MPLS Fault Management Operations, Administration, and Maintenance (OAM)
updated by
- rfc-7214 — Moving Generic Associated Channel (G-ACh) IANA Registries to a New Registry
Extracted elements (28)
Fault Management OAM suppresses alarm storms by informing client MEPs of server-layer failures before the client detects a loss of continuity, allowing operations to focus on the actual faulty elements rather than cascading alarms from dependent layers.
The clearing mechanism uses a 3.5× Refresh Timer expiration as a passive clearing method, with an active R-Flag procedure for faster clearing. The two-method design accommodates both normal fade-out and explicit notification without requiring the receiver to track sender state beyond the timer.
The L-Flag is not automatically propagated in hierarchical LSP scenarios, and the server-failure determination rules of Section 2.1.1 must be re-applied at each layer. This prevents premature triggering of protection switching at higher layers during transient conditions.
A node MUST NOT set the L-Flag until the fault has been determined to be a server-failure. A node MUST set the L-Flag if the fault has been determined to be a server-failure. During a server layer protection switching event, the L-Flag MUST NOT be set.
If the CC function is disabled for the server LSP, a node SHOULD generate AIS messages toward any clients when either the AIS or LKR indication is raised. The L-Flag is not automatically propagated during hierarchical propagation.
If the R-Flag clearing procedures are to be used, the IF_ID TLV MUST be included in the FM message. After initial transmission, the message MUST be retransmitted two more times at one-second intervals, then further retransmissions follow the Refresh Timer value.
Minimum implementation MUST support: sending AIS and LKR messages at one per second, setting the L-Flag to indicate server-failure, and receiving AIS and LKR messages with any allowed Refresh Timer value.
The AIS message SHOULD be sent as soon as the fault condition is detected, but MAY be delayed due to processing, and MAY be suppressed if protection is achieved very rapidly.
The Refresh Timer value MUST NOT be changed on successive FM messages reporting the same incident. If optional clearing procedures are not used, the default is one second; otherwise the default is 20 seconds.
When a fault is cleared, a node MUST cease sending the associated FM messages. To clear more quickly, the R-Flag is set to 1, the message is sent immediately and retransmitted two more times at one-second intervals. If a new fault occurs, the node MUST cease these retransmissions and generate new FM messages.
Fault Management messages are carried in-band of the client LSP or MS-PW using the Associated Channel Header (ACH). For LSPs other than PWs, the ACH is identified by the Generic Associated Channel Label (GAL) as defined in RFC 5586.
Fault OAM messages are generated by intermediate nodes where a client LSP is switched. When a server sub-layer fails, the intermediate node sends Fault Management messages downstream towards the LSP end point and retransmits them periodically until the condition is cleared.
The Link Down Indication (LDI) is communicated by setting the L-Flag to 1 in an AIS message. Receipt of an AIS with L-Flag set MAY be treated as equivalent to Loss of Continuity (LOC) at the client layer; the choice depends on the rate at which Continuity Check is running.
The Refresh Timer field specifies the maximum time in seconds between successive FM messages. The range is 1 to 20 seconds; the value 0 is not permitted.
Two messages are defined to communicate service disruptive conditions: Alarm Indication Signal (AIS), sent in response to detecting faults in the server sub-layer, and Lock Report (LKR), generated when a server sub-layer entity has been administratively locked.
IANA assigned Associated Channel Type 0x0058 (Fault OAM, TLV Follows=No) from the Pseudowire Associated Channel Types registry for the MPLS Fault Management channel.
RFC 6427 creates the 'MPLS Fault OAM Flag Registry' (sub-registry of MPLS OAM Parameters) with 8 flag bits (0-7), all allocated by Standards Action. Defines: bit 6 (0x2)=L-Flag, bit 7 (0x1)=R-Flag; bits 0-5 unassigned.
RFC 6427 creates the 'MPLS Fault OAM Message Type Registry' (sub-registry of MPLS OAM Parameters). Values 0-251 via Standards Action; 252-255 Experimental Use (MUST NOT be allocated). Defines: 0=Reserved, 1=AIS, 2=LKR.
RFC 6427 creates the 'MPLS Fault OAM TLV Registry' (sub-registry of MPLS OAM Parameters). Values 0-191 via Standards Action; 192-247 via Specification Required; 248-255 Experimental Use (MUST NOT be allocated). Defines: 0=Reserved, 1=Interface Identifier TLV, 2=Global Identifier.
If external MPLS traffic is mapped to an LSP via a PHP forwarding operation, it is possible to insert a GAL followed by a fault OAM message. An operator SHOULD protect against this by filtering fault OAM messages with the GAL at the top of the label stack.
Spurious fault OAM messages form a denial-of-service vector that can suppress alarms falsely. Because these messages are in a control channel, an attacker must subvert a transit node to inject them; transport networks operated as walled gardens reduce this risk.
The G-ACh can be used as a covert channel providing end-to-end connectivity that SHOULD NOT be policed by transit nodes, meaning there is no simple way to prevent traffic between consenting nodes from using this channel.
On receiving an FM message with R-Flag=0: if no matching condition exists, enter the condition and set Expiration timer to 3.5 × Refresh Timer; if condition exists, treat as refresh and reset the timer. On R-Flag=1: clear the matching condition by message type and IF_ID. Expiration timer expiry clears the condition.
Fault Management TLVs use an 8-bit Type, 8-bit Length, and variable-length Value field. The Interface Identifier TLV (Type=1, Length=0x8) carries an MPLS-TP Node Identifier (32 bits) and MPLS-TP Interface Number (32 bits) as defined in RFC 6370.
The Fault Management message format contains: 4-bit Version (currently 1), 4-bit Reserved (MUST be zero), 8-bit Message Type, 8-bit Flags, 8-bit Refresh Timer (1-20 seconds, 0 not permitted), 8-bit Total TLV Length, followed by TLVs.
The Flags field contains two defined bits: L-Flag (bit 6, hex 0x2) for Link Down Indication and R-Flag (bit 7, hex 0x1) to indicate removal of a previously sent FM condition. Reserved flags MUST be set to zero on transmission and ignored on receipt.
The Global Identifier TLV (Type=2, Length=0x4) carries the 32-bit MPLS-TP Global Identifier as defined in RFC 6370.
The MPLS Fault Management Channel is identified by ACH with Associated Channel Type 0x0058. The FM Channel MUST NOT include the ACh TLV header. The ACH consists of a 4-bit version prefix '0001', Version, Reserved, and the 0x0058 FM Channel code point in a 32-bit header.