The MPLS WG Archive[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index][Thread Index][Author Index][Subject Index] suggested clarification to section 4 of draft-ietf-mpls-lsp-ping-01
I intend this to send three parts. So far the comments are very
constructive. This is part two. Same format - diffs.
I'm offering this to the authors as my interpretation of their intent
so if I got it wrong please say so. I also may be rushing this but
hopefully I didn't say anything too stupid (not since my last response
on ATM/FR anyway:).
The next part is the multipath stuff (take two on that).
I dont' think this completes MPLS ping but I think its a step forward.
The real authors should stop me if I shouldn't be doing this.
Curtis
@@ -790,24 +889,129 @@
4.5. Non-compliant Routers
If the egress for the FEC Stack being pinged does not support MPLS
ping, then no reply will be sent, resulting in possible "false
negatives". If in "traceroute" mode, a transit LSR does not support
MPLS ping, then no reply will be forthcoming from that LSR for some
TTL, say n. The LSR originating the echo request SHOULD try sending
the echo request with TTL=n+1, n+2, ..., n+k in the hope that some
transit LSR further downstream may support MPLS echo requests and
reply. In such a case, the echo request for TTL>n MUST NOT have
Downstream Mapping TLVs, until a reply is received with a Downstream
Mapping.
+
+4.6. Operation of Traceroute Mode
+
+ In traceroute mode, either the set of one or more paths for an LSP
+ is being determined, or after a fault has been identified, the
+ approximate location of the fault is being determined.
+
+ In traceroute mode the ingress begins with a TTL=1. The test
+ traffic will reach the first LSR that decrements TTL. In the past
+ LSR have been configured to not decrement TTL. This practice is
+ discouraged, and may no longer be in use. For LSPs with more than
+ one hop that decrements TTL, when TTL is decremented the "TTL
+ Expired" processing will be initiated. The destination address in
+ the range of 127/8 will direct the packet on an MPLS ping compliant
+ router to local delivery to the 3503 UDP port. The packet will be
+ processed as defined in "4.2. Receiving an MPLS Echo Request" and
+ "4.3. Sending an MPLS Echo Reply".
+
+ As TTL is incremented by the ingress, traceroute mode test traffic
+ will reach further intermediate hops. The intended control plane
+ information is returned in Downstream Mappings. For each increment
+ of TTL another forwarding hop is exercised and further verfied by
+ the next hop downstream.
+
+ At multipath (MP) branch points more than one Downstream Mapping is
+ applicable. Multiple Downstream Mapping TLVs are returned,
+ encapsulated in a single Downstream Multipath Mapping TLV. The
+ Multipath Exercise TLVs contained in the Downstream Multipath
+ Mapping TLV provide destination addresses which can be used to
+ exercise the forwarding for specific component paths eminating from
+ the MP branch point.
+
+ At hierarchical PSC type hops, the midpoint for an LSP under test
+ is the ingress for an outer LSP that will be tested by further
+ incrementing TTL if Uniform Mode TTL ingress processing is used or
+ skipped transparently if Pipe Mode TTL ingress processing is used.
+ [***issue: The midpoint LSR needs to indicate the FEC to be added
+ to the FEC stack if Uniform Mode TTL will be used. This will be
+ the FEC of the LSP that will see the next TTL expired, at an LSR
+ that does not have knowledge of the LSPs under the stack. It would
+ appear that an extension to the Downstream Mapping TLV is needed to
+ support this as pointed out on the WG mailing list.***]
+
+ Note that FSC, LSC, and TDM hierarchical hops do not decrement TTL
+ and therefore are always transparent to MPLS ping traceroute,
+ appearing as a single hop.
+
+ When the egress of an outer LSP is reached, the normal POP
+ operation is performed. If the push was the initial push provided
+ by the ingress of the test traffic, then all lower TTLs will be one
+ and force TTL Expire processing. If the push was the result of
+ forwarding through a hierarchical PSC type hop, then the next TTL
+ will be greater than one for an outer LSP of more than one TTL
+ decrementing hop. If the egress is using Pipe Mode and outer TTL
+ was forced to a high value at the ingress, then TTL will be one.
+ If the ingress copied TTL from the inner label to the new outer
+ label, then TTL will be greater than one when the LSP is exit. If
+ so, then the outer LSP will be traced, but some hops downstream of
+ the egress of the hierarchical LSP will be skipped, including the
+ egress of the hierarchical LSP.
+
+ Similar TTL issues apply to the egress of the bottom label where
+ such a label is related to a VPN or PW. If the TTL value is
+ greater than one but additional labels are present on the stack,
+ the additional labels should be processed. The explicit null label
+ with TTL=1 should result in local delivery of the packet.
+
+ When the egress of the bottom label and bottom FEC is reached, no
+ further downstream mappings are available. The ability to extend
+ MPLS ping to indicate further forwarding, such as VPN or PW egress
+ interface or IP route can be exercised in future iterations of this
+ document or in companion documents which extend the TLV set.
+
+4.7 Operation in Ping Mode
+
+ In ping mode, one or more LSP are being continuity tested. TTL is
+ set to 255 for those LSP being tested. The packet should be
+ forwarded using normal MPLS forwarding until the egress of the
+ bottom LSP under test is reached. This LSR provides the continuity
+ verification through its MPLS ping Echo Reply.
+
+4.8 Intended Use of Traceroute and Ping Mode
+
+ Traceroute mode has two uses. It can be used prior to or after a
+ fault is detected by Ping Mode.
+
+ Traceroute mode can be used to identify the current path prior to
+ using Ping Mode and determine if multiple paths are available and
+ need to be tested. For multipath destination addresses which
+ exercise all paths can be determined.
+
+ Ping mode is used to provide a lower overhead test of forwarding.
+ If multipath is present or could arise from route change,
+ destination addresses determined using traceroute and/or randomly
+ selected addresses in the range 127/8 should be used. In the event
+ that multipath handling changes, randomly selected addresses should
+ be used in addition to destination addresses determined using
+ traceroute, rather than solely using those determined by
+ traceroute.
+
+ If ping mode failures occur, traceroute mode can be use to isolate
+ the failure usually to one of two LSR (one the sending or receiving
+ end of the hop for which forwarding stops). What action to take on
+ the detection and isolation of failure is outside the scope of this
+ document.
5. Reliable Reply Path
One of the issues that are faced with MPLS ping is to distinguish
between a failure in the forward path (the MPLS path being 'pinged')
and a failure in the return path. Note that this problem exists with
vanilla IP ping as well. In the case of MPLS ping, it is assumed
that the IP control and data planes are reliable. However, it could
be that the forwarding in the return path is via an MPLS LSP.
In this specification, we give two solutions for this problem. One
|
|