Back to EIGRP Guide Control-plane communication

EIGRP Packet Types: See How Routers Communicate

Follow EIGRP packets through neighbor discovery, route exchange, acknowledgement, and failure recovery. Learn what each packet means in a capture or troubleshooting session.

HelloUpdateQueryReplyACK

Packet Types at a Glance

The five core packet types do different jobs. Hello finds a neighbor. Update shares route information. Query and Reply help DUAL search for a path. ACK confirms reliable delivery.

DiscoveryHello
Route exchangeUpdate
Path searchQuery + Reply
Delivery proofACK
Fast memory aid: Say hello, send an update, ask a query, receive a reply, and acknowledge reliable information.

In This Lesson

  1. Read a complete packet exchange
  2. Compare the core packet types
  3. Understand Hello packets
  4. Understand Update packets
  5. Follow reliable delivery and ACKs
  6. Follow Query and Reply during failure
  7. Recognize SIA-Query and SIA-Reply
  8. Compare multicast and unicast delivery
  9. Read packet and command evidence
  10. Troubleshoot from the observed packet
  11. Complete guided packet labs
  12. Review the frequently asked questions

1. Read a Complete Packet Exchange

Follow the numbered arrows from 1 to 6. R1 and R2 first discover each other with Hello packets, then exchange their initial routing information with reliable Update and ACK packets.

EIGRP neighborship process between R1 and R2 showing Hello discovery, initial Update exchange, acknowledgements, and the adjacency becoming active
How to read it: the arrows show normal neighbor formation from discovery through the initial topology exchange. Each ACK confirms that a reliable Update arrived.
Important distinction: Query and Reply are not part of normal neighbor startup. EIGRP uses them later when a route changes and DUAL must search for another path.

2. Compare the Core Packet Types

Start with purpose, then add delivery behavior. This prevents the common mistake of assuming that every EIGRP packet is multicast or reliable.

PacketMain jobReliable?Typical timing
HelloDiscover and maintain neighborsNoSent periodically
UpdateShare route and metric informationYes when reliable delivery is usedInitial synchronization or route change
QueryAsk whether a neighbor has another pathYesNo feasible successor is available
ReplyAnswer a QueryYesAfter checking local topology and downstream answers
ACKConfirm receipt of a reliable packetNo acknowledgement of the ACKAfter Update, Query, or Reply delivery

3. Understand Hello Packets

Hello packets discover neighbors and keep them alive. IPv4 EIGRP normally sends them to multicast address 224.0.0.10. EIGRP for IPv6 uses FF02::A.

Common LAN defaults

Hello interval: 5 seconds
Hold time:     15 seconds
IPv4 group:    224.0.0.10
IP protocol:   88

Neighbor requirements

Compatible primary subnet
Matching EIGRP AS
Matching K-values
Matching authentication
Interface is not passive
  • Hold time: A router advertises how long its neighbor should keep the relationship without another valid Hello.
  • Timers may differ: EIGRP Hello and hold timers do not have to match between neighbors.
  • Passive interface: The connected network can still be advertised, but no Hello is sent and no neighbor forms there.
Capture clue: Hellos arriving in only one direction suggest a return-path, ACL, switch, or interface problem—not necessarily an EIGRP metric problem.

4. Understand Update Packets

New neighbors use Update packets to synchronize topology information. After that, EIGRP sends partial, bounded updates: only changed information goes to routers that need it.

Neighbor forms

Both routers pass the Hello checks and create a neighbor entry.

Topology sync

Update packets carry destinations, metric components, and route attributes.

Stable state

Later updates are triggered by changes instead of a full periodic table refresh.

SituationWhat the Update carriesOperational meaning
New neighborInitial topology informationThe peers build a shared routing view
Metric changesChanged destination and metric dataDUAL checks whether the best path changed
Route withdrawalDestination marked unreachableThe old path cannot remain the successor

5. Follow Reliable Delivery and ACKs

EIGRP's Reliable Transport Protocol (RTP) tracks selected packets with sequence numbers. The receiver sends an ACK, and the sender can retransmit if the acknowledgement does not arrive.

1

Send

R1 sends a reliable packet with a sequence number.

2

Acknowledge

R2 sends an ACK for that sequence number.

3

Retransmit if needed

R1 retries if the expected ACK is missing.

ACK detail: An EIGRP ACK uses the Hello opcode with no route payload. It confirms one reliable packet and is not acknowledged again.

In show ip eigrp neighbors, SRTT is the measured smooth round-trip time and RTO is the retransmission timeout. A queue count that stays above zero is more concerning than a brief change during normal convergence.

6. Follow Query and Reply During a Failure

When a successor fails, DUAL first looks for a feasible successor. If none exists, the route becomes Active and the router asks eligible neighbors for help.

1

Successor fails

The current best path is no longer usable.

2

Check backup

DUAL looks for a feasible successor.

3

Send Query

No safe backup exists, so neighbors are asked.

4

Collect Replies

DUAL waits for every required answer before finishing.

Neighbor resultReply meaningWhat happens next
Has another pathReturns reachable metric informationDUAL evaluates the new path
Has no pathReports the destination unreachableThe search continues through other required neighbors
Must ask downstreamKeeps its own route ActiveReply waits until downstream answers arrive

7. Recognize SIA-Query and SIA-Reply

The traditional list has five core packets. Modern EIGRP also uses SIA-Query and SIA-Reply packets to show that a neighbor is still working on an outstanding route query.

  • SIA-Query: Asks whether the neighbor is still processing the original Query.
  • SIA-Reply: Confirms that the search is still in progress, helping preserve the neighbor relationship.
  • Still investigate: Repeated SIA behavior can point to excessive query scope, packet loss, slow links, CPU pressure, or an unstable design.
Design lesson: Summaries and EIGRP stub routers limit how far Queries travel. A smaller query domain usually converges more predictably.

8. Compare Multicast and Unicast Delivery

Do not identify a packet only by its destination address. EIGRP can use multicast for a group of neighbors and unicast for one specific peer or retransmission.

DeliveryTypical useTroubleshooting clue
Multicast 224.0.0.10IPv4 Hellos and efficient delivery to several EIGRP peersCheck multicast filtering and interface participation
IPv6 multicast FF02::AEIGRP for IPv6 neighbor communicationCheck link-local addressing and IPv6 multicast handling
UnicastACKs, selected exchanges, or retransmission to one neighborCheck two-way unicast reachability and interface errors

EIGRP is carried directly in IP protocol 88. It does not use TCP or UDP ports, so a firewall or ACL must permit the IP protocol itself.

9. Read Packet and Command Evidence

Begin with show commands. Use a controlled debug or packet capture only when the neighbor and topology output cannot explain the problem.

Safe first checks

show ip eigrp neighbors
show ip eigrp interfaces detail
show ip eigrp topology
show ip protocols
show interfaces counters errors

Focused lab debug

debug eigrp packets hello
debug eigrp packets update
debug eigrp packets query
debug eigrp packets reply

undebug all
FieldMeaningHealthy pattern
HoldTime before the peer is declared downCounts down, then refreshes
UptimeAge of the neighbor relationshipIncreases without frequent resets
SRTTSmoothed reliable-packet round-trip timeReasonable and stable for the link
RTOTime before retransmittingTracks link conditions without repeated expiry
Q countReliable packets waiting to be sentNormally zero
Seq NumLast reliable sequence receivedChanges as reliable traffic is exchanged
Production safety: Debug output can be heavy. Use it briefly, filter when supported, send output to a controlled destination, and always finish with undebug all.

10. Troubleshoot from the Observed Packet

The packet pattern tells you how far EIGRP progressed. Match the observation to the next evidence to collect.

Observed patternLikely layerCheck next
No HellosEIGRP not enabled, passive interface, ACL, or link issueInterface state and EIGRP interface list
Hellos seen, no neighborAS, K-value, authentication, or subnet mismatchshow ip protocols and both interface configurations
Updates retransmittedMissing ACK, loss, congestion, or one-way filteringQ count, RTO, interface errors, and both traffic directions
Many QueriesNo feasible successor or query domain too largeTopology table, summaries, stubs, and failure location
Query without final ReplyDownstream delay, packet loss, or overloaded neighborActive topology entries and the named outstanding peer
Neighbor repeatedly resetsHold timer expiry or reliable transport failureLogs, link stability, CPU, and packet counters
  1. State exactly which packet is present, missing, or repeating.
  2. Check the neighbor and topology state at the same time.
  3. Compare both ends; one-way evidence is not enough.
  4. Repair one confirmed cause and repeat the same capture or show command.

11. Guided Packet Labs

Use two or three routers in an isolated EIGRP AS 100 lab. Restore the healthy baseline after each task.

Task 1: Observe Neighbor Startup

  1. Enable EIGRP on one link.
  2. Capture or debug Hello packets briefly.
  3. Enable the second router.
  4. Record when the neighbor appears and when Updates begin.
  5. Stop the debug.

Expected result: You separate Hello discovery from topology synchronization.

Task 2: Find an AS Mismatch

  1. Change one router from AS 100 to AS 200.
  2. Confirm Hellos exist but no neighbor forms.
  3. Compare show ip protocols on both routers.
  4. Restore AS 100.
  5. Confirm the exchange completes.

Expected result: You avoid blaming the physical link when EIGRP parameters disagree.

Task 3: Trigger a Query

  1. Build a three-router path with no feasible successor.
  2. Remove the successor link.
  3. Watch the route become Active.
  4. Identify the Query direction and Reply source.
  5. Confirm the route returns to Passive.

Expected result: You connect DUAL state to Query and Reply packets.

Task 4: Investigate Reliability

  1. Record SRTT, RTO, and Q count in the healthy state.
  2. Introduce controlled loss only if your lab supports it.
  3. Watch for retransmission or queue changes.
  4. Restore the link.
  5. Confirm Q count returns to zero.

Expected result: You use neighbor-table evidence before enabling a broad debug.

12. Frequently Asked Questions

Are there five or seven EIGRP packet types?

Hello, Update, Query, Reply, and ACK are the five traditional core types. Modern EIGRP also uses SIA-Query and SIA-Reply to manage long-running route searches.

Does EIGRP use TCP or UDP?

No. EIGRP is carried directly in IP protocol 88. Its own Reliable Transport Protocol handles acknowledgements and retransmissions.

Are Hello packets acknowledged?

Normal periodic Hello packets are not acknowledged. An EIGRP ACK uses the Hello opcode, but it contains an acknowledgement number and no routing payload.

When does EIGRP send a Query?

A route sends Queries when its successor is lost and no feasible successor can be used immediately. DUAL asks eligible neighbors for another path.

Does every route change trigger a network-wide Query?

No. A feasible successor can avoid a Query, and summaries or EIGRP stub boundaries can stop Queries from spreading unnecessarily.

Why can an Update be unicast?

EIGRP can target one neighbor during selected exchanges or retransmissions. Multicast is efficient for several peers, but unicast is useful when only one peer needs the packet.

What does a nonzero Q count mean?

It means reliable packets are waiting in the neighbor queue. A brief value can occur during change; a value that stays above zero can indicate loss, congestion, or acknowledgement trouble.