I85PREP
⌘ K
JNCIP-SP study materials
JNCIP-SP study resources

RSVP-TE LSP Down: Admin Group Versus Bandwidth

JNCIP-SP · Put your knowledge to work in context.

All case studies

RSVP-TE LSP Down: Admin Group Versus Bandwidth

PE-1 needs an RSVP-TE LSP to PE-2 to carry a 500 Mbps VoIP trunk. The LSP is configured with bandwidth 500m and admin-group include-any gold. Two paths exist. Path A is PE-1 to P-1 to P-2 to PE-2: every link carries admin-group gold, but the P-1–P-2 link (10 Gbps) already has 9.6 Gbps reserved by three other LSPs. Path B is PE-1 to P-3 to PE-2 with 8 Gbps free on every link, but the P-3–PE-2 link is not tagged gold because it is a new link that was never added to the admin group. The LSP stays down; show mpls lsp name to-PE2 extensive shows the path computation failing, and show rsvp session shows no session. The VoIP trunk is currently riding a best-effort LDP path with visible jitter. The LSPs already on P-1–P-2 belong to a business customer with a 99.99% SLA and must not be disrupted.

1. Both constraints are failing candidates. How do you determine which one CSPF actually rejected, and what does each command tell you?

Read the worked answer

Do not guess between admin group and bandwidth - a misdiagnosis here leads directly to a disruptive fix. Read the evidence in this order:

  1. show mpls lsp name to-PE2 extensive - the path history reports the CSPF result and the reason (for example no route found versus an admin-group mismatch). This tells you whether the failure is at computation or at reservation.
  2. show ted database 10.255.0.2 extensive - the TED entry per link carries the admin group bit masks and the available bandwidth per priority 0–7. On the P-1–P-2 link you will see gold set and only ~400 Mbps available at the requested priority; on P-3–PE-2 you will see plenty of bandwidth and gold absent.
  3. show rsvp interface detail on P-1 - confirms reservation data independently of the TED and shows the interface subscription percentage. If subscription is 100 percent and 9.6 G of 10 G is reserved, no further reservation is possible regardless of the TED.
  4. show ted database link count between P-3 and PE-2 - verifies the new link is actually flooded into the TED. A link that exists in IS-IS but is excluded from traffic engineering will not appear at all, which changes the answer.

In this case both constraints bite: Path A passes admin group and fails bandwidth, Path B passes bandwidth and fails admin group, so no path satisfies both.

2. Rank the remediation options and justify the choice given the 99.99% SLA on the existing LSPs.

Read the worked answer

Options, best to worst:

  1. Tag the new P-3–PE-2 link with gold and re-signal the LSP. This restores the design intent - a gold-only topology with real capacity - and is non-disruptive. Verify first with show ted database 10.255.0.2 extensive that the link is flooded and has 500 Mbps available at the LSP's setup priority, then set protocols mpls interface ge-x/y/z admin-group gold, commit, and let CSPF re-run. Confirm the new path with show mpls lsp name to-PE2 extensive.
  2. Relax to admin-group include-any with no group, or remove the admin-group statement. CSPF will then pick Path B on bandwidth. It works, but it silently places a latency-sensitive trunk on a link outside the engineered gold plane, and the same LSP may later move onto an undesired link. Acceptable only as a temporary measure with a documented follow-up.
  3. Raise RSVP subscription above 100 percent on P-1–P-2. Admission control would succeed on paper while the link is already 96 percent reserved. Rejected: oversubscription of a link carrying an SLA customer.
  4. Preempt with priority 2 2. Preemption would tear down a customer's 99.99% LSP to make room for voice. Rejected outright.

Choose option 1, with option 2 only as a same-day stopgap if the new link cannot be tagged inside the window.