First, the essential ideas
- EtherChannel and port-channel
- EtherChannel groups compatible physical links into one logical interface called a port-channel. A member is one physical link in that group; a configured member is not necessarily successfully bundled.
- LACP
- Link Aggregation Control Protocol negotiates compatible aggregation with a peer. Active initiates negotiation; passive responds. Active/active or active/passive can form; passive/passive does not initiate.
- Layer 2 versus Layer 3 channel
- A Layer 2 port-channel carries Ethernet VLAN traffic as access or trunk. A Layer 3 port-channel uses no switchport and has an IP address on the logical interface, not separate addresses on its members.
- Consistency
- Member attributes must be compatible, including speed, duplex and Layer 2/Layer 3 mode. Layer 2 members must also agree on access VLAN or trunk/native/allowed settings.
- Hash and flow
- A hash selects a member from packet-header fields using a platform's distribution method. Traffic with the same relevant fields commonly stays on one member; one conversation does not automatically use the sum of link speeds.
- Bundled, suspended and standalone
- Bundled means participating in the logical channel. Suspended or standalone states mean the port is not contributing as intended. Read the command's own legend rather than equating every physical up state with a healthy bundle.
- Redundancy boundary
- A remaining eligible member can preserve logical connectivity after one member fails, subject to configuration such as minimum links. It does not protect against failure of both peer switches or every shared dependency.
- Console and baseline
- Console directly manages a named lab node. The baseline must include member participation, LACP partner evidence and the same traffic test before a controlled member shutdown.
Think of a port-channel as one logical interface backed by several compatible links. First prove those links are actually participating, then test the service it carries. The two lab pairs deliberately separate VLAN carriage from routed addressing.
See the idea
Two cables, one logical routed link
Both members must join LACP. The /30 address belongs to Port-channel2, not to either cable.
Verify both members
RSW1 and RSW2 use two physical cables in one LACP aggregate. Both members must show actual participation.
What this model leaves out: The two member markers represent cables, not extra devices. Only the routed pair is shown; the separate Layer 2 pair is outside this figure. No bandwidth measurement is implied.
Full visual explanation and references
RSW1 and RSW2 share two parallel member cables in Port-channel2. Member A is Gi0/0 and member B is Gi0/1. Both endpoints retain their logical /30 addresses.
The two member markers represent cables, not extra devices. Only the routed pair is shown; the separate Layer 2 pair is outside this figure. No bandwidth measurement is implied.
- Verify both members
RSW1 and RSW2 use two physical cables in one LACP aggregate. Both members must show actual participation.
www.cisco.com - Lose one member
Shutting RSW1 Gi0/0 removes member A. Member B can keep the routed logical peer reachable; exact loss is unmeasured.
www.cisco.com - Restore participation
Re-enable Gi0/0, verify LACP membership and repeat the /30 peer test. An enabled cable alone is not a recovered member.
www.cisco.com
One idea at a time
Start with something familiar.
Draw two parallel cables between SW1 and SW2 and write Port-channel1 around the pair. Separately draw RSW1 and RSW2 with Port-channel2. Do not draw a cable between the two experiments.
Why learn this?
Two green cables do not prove a working aggregate. A mismatch may leave one member suspended while traffic still works through the other, hiding lost capacity and redundancy.
An everyday comparison
Several compatible checkout lanes serve one queue, but one purchase is assigned to a lane rather than split item-by-item across every cashier.
What happens in a network
The shared queue resembles a logical port-channel and the lanes resemble members. A selection method distributes different traffic across available members.
Where the comparison stopsEtherChannel uses header-based selection, not a human queue manager. The analogy does not predict exact bandwidth, perfect fairness or packet timing.
Follow one worked example
- Agree on compatible member attributes.
Both access members use VLAN 10, while both routed members use no switchport and no individual IP address.
Why: Different interface roles cannot safely form one consistent logical service.
- Establish LACP participation.
Active peers initiate negotiation. Summary flags and neighbor records show whether each member actually joined.
Why: A configured channel-group line is intent, not proof of successful bundling.
- Test a member failure and restore it.
The remaining eligible member should preserve the logical peer test. The restored member must renegotiate before being counted again.
Why: Redundancy is a behavior to verify within its limits, not a promise based on cable count.
Your first small task
On paper, compare active/active, active/passive and passive/passive. Then place a /30 IP address on either the physical members or Port-channel2 and explain the correct location for the routed design.
Show a hint
Separate negotiation role from forwarding role; passive does not mean a successfully bundled member is reserved only for failure.
Check the expected result
The first two mode pairs can negotiate; passive/passive does not initiate. The routed /30 address belongs on Port-channel2.
Pause and explain it in your own words
Will one file transfer necessarily run at twice one member's speed?
Show a hint
Consider how the same flow's headers select a member.
Show the explanation
No. A flow commonly maps to one member under the selected hash. Multiple flows may use aggregate capacity, but this virtual lab establishes neither doubled flow speed nor hardware throughput.
Short answer
What should you understand today?
LACP combines compatible links into a logical port-channel when at least one side initiates negotiation. A Layer 2 channel carries VLAN traffic; a Layer 3 channel has its IP address on the logical interface with unaddressed routed members. Verify both partner negotiation and member participation. Redundancy can preserve service after one link fails, but one flow does not automatically gain all members' combined bandwidth.
Before you begin
Learning targets
- Understand access VLANs, SVIs and routed physical ports.
- Use four fresh IOSvL2 nodes under an existing CML 2.10 entitlement.
- Check actual image support for LACP and routed port-channels before any experiment.
- Predict active/active, active/passive and passive/passive negotiation.
- Configure both access and routed LACP channels.
- Verify partner identity and member bundling instead of relying on link-up state.
- Diagnose consistency problems without forcing static aggregation.
- Test and restore one member in each isolated channel type.
Words you will meet
Open a word when you need its meaning. You do not need to memorize the whole list before reading.
EtherChannel and port-channel
EtherChannel groups compatible physical links into one logical interface called a port-channel. A member is one physical link in that group; a configured member is not necessarily successfully bundled.
LACP
Link Aggregation Control Protocol negotiates compatible aggregation with a peer. Active initiates negotiation; passive responds. Active/active or active/passive can form; passive/passive does not initiate.
Layer 2 versus Layer 3 channel
A Layer 2 port-channel carries Ethernet VLAN traffic as access or trunk. A Layer 3 port-channel uses no switchport and has an IP address on the logical interface, not separate addresses on its members.
Consistency
Member attributes must be compatible, including speed, duplex and Layer 2/Layer 3 mode. Layer 2 members must also agree on access VLAN or trunk/native/allowed settings.
Hash and flow
A hash selects a member from packet-header fields using a platform's distribution method. Traffic with the same relevant fields commonly stays on one member; one conversation does not automatically use the sum of link speeds.
Bundled, suspended and standalone
Bundled means participating in the logical channel. Suspended or standalone states mean the port is not contributing as intended. Read the command's own legend rather than equating every physical up state with a healthy bundle.
Redundancy boundary
A remaining eligible member can preserve logical connectivity after one member fails, subject to configuration such as minimum links. It does not protect against failure of both peer switches or every shared dependency.
Console and baseline
Console directly manages a named lab node. The baseline must include member participation, LACP partner evidence and the same traffic test before a controlled member shutdown.
Negotiation is different from load sharing
LACP active initiates while passive responds; at least one side must initiate. Both active and passive members can forward after successful aggregation. PAgP uses different mode names and is not LACP. Static mode on suppresses negotiation and must not be used to hide an unexplained compatibility failure.
- Passive does not mean standby-only forwarding.
- A peer's channel number can differ; compatibility and partner association matter, not matching labels alone.
Layer 2 members carry one consistent VLAN service
An access bundle needs the same access VLAN on its members. A trunk bundle needs compatible trunk encapsulation, native VLAN and allowed lists. Speed and duplex must also be compatible. Once grouped, make intended service changes consistently through the logical interface according to platform behavior and verify all members.
- Layer 2 spanning tree treats the channel as a logical link.
- A successful ping through one member can hide failure of the other.
Layer 3 addressing belongs to the aggregate
For the routed design, create Port-channel2 as a Layer 3 interface, assign its /30 address, and join compatible unaddressed routed members. Do not put one subnet address on each member or retain access/trunk membership. The two ends use distinct addresses in the same directly connected subnet.
- Use fresh nodes rather than converting the working Layer 2 pair mid-experiment.
- If routed channel support is absent, stop this branch and label it unperformed.
Read participation and distribution separately
show etherchannel summary includes a legend for logical type, use and member participation. On common IOS output S denotes Layer 2, R Layer 3, U in use and P a bundled member, but read the displayed legend rather than guessing. show lacp neighbor identifies partner negotiation. Distribution uses platform-selected header fields and may be uneven.
- A suspended or standalone member needs investigation even if the logical peer responds.
- CML's virtual performance limits make this a configuration exercise, not a bandwidth benchmark.
Recover a member without forcing a loop
Compare peer cabling, speed/duplex, Layer 2/Layer 3 role, VLAN attributes and negotiation modes when bundling fails. Preserve relevant error output. Correct the proven mismatch under an approved plan; do not force on mode, disable spanning tree or remove a live member from aggregation without considering loops.
- No zero-loss or exact convergence-time guarantee is made.
- v1.1 2.4 and v2.0 2.1.c include both Layer 2 and Layer 3 LACP, so both practical branches are explicit.
Real-world walkthrough
The bundle works, but one member never joined
A hypothetical uplink passes a basic ping, yet one physical member shows suspended state after a configuration change. The operator is tempted to ignore it because users still have some connectivity. The missing member means the intended redundancy and aggregate capacity are not established.
- Record both logical and individual member flags with the command's legend.
- Read LACP partner evidence and trace each cable to the intended same peer.
- Compare member role, VLAN/trunk parameters, speed and duplex against the recorded baseline.
- Correct only the identified inconsistency; do not bypass negotiation with static mode on.
- Retest the peer and verify both members actually participate before closing the change.
Guided GNS3 lab
Separate access and routed LACP experiments
Configure both required LACP forms and test one member's failure in each. All expected results are predictions; no author runtime, loss measurement or throughput measurement is claimed.
Addressing plan
| Device | Interface | Address | Purpose |
|---|---|---|---|
| SW1 | Vlan10 | 192.0.2.1/24 | L2 test source |
| SW2 | Vlan10 | 192.0.2.2/24 | L2 test peer |
| RSW1 | Port-channel2 | 203.0.113.1/30 | L3 aggregate source |
| RSW2 | Port-channel2 | 203.0.113.2/30 | L3 aggregate peer |
Set up the lab
- Create four fresh entitled CML nodes and the exact four cables. Keep the pairs isolated from one another and from external connectors.
- Before configuration, open each node's Console, use enable, then show version and show running-config. Record release and confirm no pre-existing member IPs, channel groups or minimum-links policy; these inspection commands do not modify state.
- Preserve the starting lab. Do not use routed configuration on an image that rejects no switchport or Port-channel2 addressing; record unsupported behavior rather than substituting invented output.
- Paper alternative: compare configuration roles and predict member states. It does not satisfy the practical Layer 3 branch when no supported runtime is available.
Build SW1 Layer 2 LACP
In CML, select SW1, open its Console tab and connect before typing. These two fresh ports connect only to the matching peer. Configure while administratively down, then enable them after consistent membership is set; do not force mode on.
enable
configure terminal
vlan 10
exit
interface range GigabitEthernet0/0 - 1
shutdown
switchport mode access
switchport access vlan 10
channel-group 1 mode active
no shutdown
endWhat each command means
enable- Enter privileged EXEC on the named lab node; stop if authorization is unavailable.
configure terminal- Enter global configuration mode from privileged EXEC; changes apply as accepted.
vlan 10- Create the single VLAN carried by this access bundle.
exit- Return to global configuration.
interface range GigabitEthernet0/0 - 1- Select exactly the two parallel cables to the same peer.
shutdown- Keep the selected members disabled during construction.
switchport mode access- Use consistent Layer 2 access behavior on both members.
switchport access vlan 10- Assign both members to the same VLAN.
channel-group 1 mode active- Join logical Port-channel1 using LACP initiation.
no shutdown- Enable both prepared member links.
end- Return to privileged EXEC without saving or reversing the configuration.
- What you should see
- Port-channel1 is configured as an LACP Layer 2 access bundle.
- Why this step matters
- Configuration is not participation; later summary and partner checks must show both members actually bundled.
Build SW2 Layer 2 LACP
In CML, select SW2, open its Console tab and connect before typing. These two fresh ports connect only to the matching peer. Configure while administratively down, then enable them after consistent membership is set; do not force mode on.
enable
configure terminal
vlan 10
exit
interface range GigabitEthernet0/0 - 1
shutdown
switchport mode access
switchport access vlan 10
channel-group 1 mode active
no shutdown
endWhat each command means
enable- Enter privileged EXEC on the named lab node; stop if authorization is unavailable.
configure terminal- Enter global configuration mode from privileged EXEC; changes apply as accepted.
vlan 10- Create the single VLAN carried by this access bundle.
exit- Return to global configuration.
interface range GigabitEthernet0/0 - 1- Select exactly the two parallel cables to the same peer.
shutdown- Keep the selected members disabled during construction.
switchport mode access- Use consistent Layer 2 access behavior on both members.
switchport access vlan 10- Assign both members to the same VLAN.
channel-group 1 mode active- Join logical Port-channel1 using LACP initiation.
no shutdown- Enable both prepared member links.
end- Return to privileged EXEC without saving or reversing the configuration.
- What you should see
- Port-channel1 is configured as an LACP Layer 2 access bundle.
- Why this step matters
- Configuration is not participation; later summary and partner checks must show both members actually bundled.
Address SW1 VLAN test
In CML, select SW1, open its Console tab and connect before typing. The IP address goes on Vlan10, not the Layer 2 port-channel members. Inspect the bundle after the other side is configured.
configure terminal
interface Vlan10
ip address 192.0.2.1 255.255.255.0
no shutdown
end
show etherchannel summary
show lacp neighborWhat each command means
configure terminal- Enter global configuration mode from privileged EXEC; changes apply as accepted.
interface Vlan10- Select the Layer 2 bundle's local IP test attachment.
ip address 192.0.2.1 255.255.255.0- Assign this switch's distinct local /24 test address.
no shutdown- Enable the SVI subject to active VLAN membership.
end- Return to privileged EXEC without saving or reversing the configuration.
show etherchannel summary- Read port-channel type/use and each member's participation flags.
show lacp neighbor- Record the actual LACP partner information.
- What you should see
- Vlan10 can become operational through the bundled links.
- Why this step matters
- The SVI provides the endpoint test while the Layer 2 channel remains an Ethernet interface.
Address SW2 VLAN test
In CML, select SW2, open its Console tab and connect before typing. The IP address goes on Vlan10, not the Layer 2 port-channel members. Inspect the bundle after the other side is configured.
configure terminal
interface Vlan10
ip address 192.0.2.2 255.255.255.0
no shutdown
end
show etherchannel summary
show lacp neighborWhat each command means
configure terminal- Enter global configuration mode from privileged EXEC; changes apply as accepted.
interface Vlan10- Select the Layer 2 bundle's local IP test attachment.
ip address 192.0.2.2 255.255.255.0- Assign this switch's distinct local /24 test address.
no shutdown- Enable the SVI subject to active VLAN membership.
end- Return to privileged EXEC without saving or reversing the configuration.
show etherchannel summary- Read port-channel type/use and each member's participation flags.
show lacp neighbor- Record the actual LACP partner information.
- What you should see
- Vlan10 can become operational through the bundled links.
- Why this step matters
- The SVI provides the endpoint test while the Layer 2 channel remains an Ethernet interface.
Baseline the access bundle
In CML, select SW1, open its Console tab and connect before typing. Require both Gi0/0 and Gi0/1 bundled, not merely connected. Stop if either is suspended or standalone.
show etherchannel summary
show lacp neighbor
ping 192.0.2.2What each command means
show etherchannel summary- Verify the Layer 2 logical interface and both participating members.
show lacp neighbor- Confirm the intended peer on each member.
ping 192.0.2.2- Test the VLAN 10 SVI peer through the aggregate.
- What you should see
- Expect an in-use Layer 2 channel, both bundled members and peer replies.
- Why this step matters
- This proves the starting member state before testing redundancy.
Build RSW1 routed LACP
In CML, select RSW1, open its Console tab and connect before typing. Use a separate fresh node pair, never convert the live Layer 2 exercise. Inspect the fresh member configuration first in setup; members must have no IP addresses.
enable
configure terminal
interface Port-channel2
no switchport
ip address 203.0.113.1 255.255.255.252
exit
interface range GigabitEthernet0/0 - 1
shutdown
no switchport
channel-group 2 mode active
no shutdown
endWhat each command means
enable- Enter privileged EXEC on the named lab node; stop if authorization is unavailable.
configure terminal- Enter global configuration mode from privileged EXEC; changes apply as accepted.
interface Port-channel2- Create the logical routed aggregate before assigning members.
no switchport- Make the logical port-channel a Layer 3 interface.
ip address 203.0.113.1 255.255.255.252- Assign the /30 address to the logical aggregate only.
exit- Return to global mode for physical member selection.
interface range GigabitEthernet0/0 - 1- Select the two cables to this routed peer.
shutdown- Hold both members down during consistent construction.
no switchport- Make the fresh unaddressed members Layer 3 interfaces.
channel-group 2 mode active- Bind both members to Port-channel2 with LACP.
no shutdown- Enable the prepared physical members.
end- Return to privileged EXEC without saving or reversing the configuration.
- What you should see
- Port-channel2 owns the routed address; members remain unaddressed.
- Why this step matters
- The logical interface is the Layer 3 attachment, and the separate pair avoids mixing access VLANs with routed members.
Build RSW2 routed LACP
In CML, select RSW2, open its Console tab and connect before typing. Use a separate fresh node pair, never convert the live Layer 2 exercise. Inspect the fresh member configuration first in setup; members must have no IP addresses.
enable
configure terminal
interface Port-channel2
no switchport
ip address 203.0.113.2 255.255.255.252
exit
interface range GigabitEthernet0/0 - 1
shutdown
no switchport
channel-group 2 mode active
no shutdown
endWhat each command means
enable- Enter privileged EXEC on the named lab node; stop if authorization is unavailable.
configure terminal- Enter global configuration mode from privileged EXEC; changes apply as accepted.
interface Port-channel2- Create the logical routed aggregate before assigning members.
no switchport- Make the logical port-channel a Layer 3 interface.
ip address 203.0.113.2 255.255.255.252- Assign the /30 address to the logical aggregate only.
exit- Return to global mode for physical member selection.
interface range GigabitEthernet0/0 - 1- Select the two cables to this routed peer.
shutdown- Hold both members down during consistent construction.
no switchport- Make the fresh unaddressed members Layer 3 interfaces.
channel-group 2 mode active- Bind both members to Port-channel2 with LACP.
no shutdown- Enable the prepared physical members.
end- Return to privileged EXEC without saving or reversing the configuration.
- What you should see
- Port-channel2 owns the routed address; members remain unaddressed.
- Why this step matters
- The logical interface is the Layer 3 attachment, and the separate pair avoids mixing access VLANs with routed members.
Baseline the routed bundle
In CML, select RSW1, open its Console tab and connect before typing. Inspect the logical interface and unaddressed members. A routed bundle must not still be a Layer 2 access channel.
show etherchannel summary
show lacp neighbor
show ip interface brief
ping 203.0.113.2What each command means
show etherchannel summary- Verify Layer 3 type, logical use and both member states.
show lacp neighbor- Confirm the routed pair's actual partner.
show ip interface brief- Verify the /30 address belongs to Port-channel2.
ping 203.0.113.2- Test the directly connected routed peer.
- What you should see
- Expect a usable Layer 3 channel with both members bundled and the peer replying.
- Why this step matters
- The routed branch establishes its own baseline rather than borrowing Layer 2 evidence.
Fail one Layer 2 member
In CML, select SW1, open its Console tab and connect before typing. Proceed only after both members are bundled and the relevant peer test works. Shut Gi0/0 only; leave Gi0/1 and the logical port-channel enabled.
configure terminal
interface GigabitEthernet0/0
shutdown
end
show etherchannel summaryWhat each command means
configure terminal- Enter global configuration mode from privileged EXEC; changes apply as accepted.
interface GigabitEthernet0/0- Select exactly one member of this node's bundle.
shutdown- Simulate a single-member failure, not a whole-channel failure.
end- Return to privileged EXEC without saving or reversing the configuration.
show etherchannel summary- Observe the failed member and remaining participating member.
- What you should see
- Expect one member removed from service while the other can keep the logical interface usable.
- Why this step matters
- This tests one redundancy boundary without creating a loop or forcing inconsistent aggregation.
Test the access bundle on one member
In CML, select SW2, open its Console tab and connect before typing. Leave the fault in place while inspecting the peer and repeating the original destination in the reverse direction.
show etherchannel summary
ping 192.0.2.1What each command means
show etherchannel summary- Confirm only the remaining healthy member participates.
ping 192.0.2.1- Test SW1's SVI through the remaining member.
- What you should see
- Expect continued peer connectivity after convergence with one member active.
- Why this step matters
- The test covers a single member failure, not a switch failure or a zero-loss guarantee.
Restore the Layer 2 member
In CML, select SW1, open its Console tab and connect before typing. Restore only the member disabled by this experiment. Wait for actual LACP negotiation before counting the member as recovered.
configure terminal
interface GigabitEthernet0/0
no shutdown
end
show etherchannel summary
show lacp neighbor
ping 192.0.2.2What each command means
configure terminal- Enter global configuration mode from privileged EXEC; changes apply as accepted.
interface GigabitEthernet0/0- Select the same failed member.
no shutdown- Return it to the enabled baseline.
end- Return to privileged EXEC without saving or reversing the configuration.
show etherchannel summary- Verify both members are again bundled using the displayed legend.
show lacp neighbor- Confirm partner negotiation on the restored links.
ping 192.0.2.2- Repeat the original peer test through the restored logical interface.
- What you should see
- Expect both participating members and successful peer replies after negotiation.
- Why this step matters
- An administratively enabled cable is not enough; membership and actual traffic must also recover.
Fail one routed member
In CML, select RSW1, open its Console tab and connect before typing. Proceed only after both members are bundled and the relevant peer test works. Shut Gi0/0 only; leave Gi0/1 and the logical port-channel enabled.
configure terminal
interface GigabitEthernet0/0
shutdown
end
show etherchannel summaryWhat each command means
configure terminal- Enter global configuration mode from privileged EXEC; changes apply as accepted.
interface GigabitEthernet0/0- Select exactly one member of this node's bundle.
shutdown- Simulate a single-member failure, not a whole-channel failure.
end- Return to privileged EXEC without saving or reversing the configuration.
show etherchannel summary- Observe the failed member and remaining participating member.
- What you should see
- Expect one member removed from service while the other can keep the logical interface usable.
- Why this step matters
- This tests one redundancy boundary without creating a loop or forcing inconsistent aggregation.
Test the routed bundle on one member
In CML, select RSW2, open its Console tab and connect before typing. Keep the Layer 3 fault isolated from the restored access pair. Inspect logical state and test its directly connected peer.
show etherchannel summary
ping 203.0.113.1What each command means
show etherchannel summary- Read remaining routed-member participation.
ping 203.0.113.1- Test RSW1's logical /30 endpoint over the surviving member.
- What you should see
- Expect the routed peer to remain reachable after negotiation/convergence.
- Why this step matters
- The Layer 3 aggregate address stays logical; it is not moved manually to the surviving physical port.
Restore the routed member
In CML, select RSW1, open its Console tab and connect before typing. Restore only the member disabled by this experiment. Wait for actual LACP negotiation before counting the member as recovered.
configure terminal
interface GigabitEthernet0/0
no shutdown
end
show etherchannel summary
show lacp neighbor
ping 203.0.113.2What each command means
configure terminal- Enter global configuration mode from privileged EXEC; changes apply as accepted.
interface GigabitEthernet0/0- Select the same failed member.
no shutdown- Return it to the enabled baseline.
end- Return to privileged EXEC without saving or reversing the configuration.
show etherchannel summary- Verify both members are again bundled using the displayed legend.
show lacp neighbor- Confirm partner negotiation on the restored links.
ping 203.0.113.2- Repeat the original peer test through the restored logical interface.
- What you should see
- Expect both participating members and successful peer replies after negotiation.
- Why this step matters
- An administratively enabled cable is not enough; membership and actual traffic must also recover.
Verify
- Require two actual bundled members and correct partners on each pair.
- Check SVI addressing for L2 and logical port-channel addressing for L3.
- Record each pair's own successful baseline before its fault.
- Retest while one member is down and note actual loss rather than promising none.
- After restoration, confirm both members rejoin and repeat the original peer test.
Troubleshoot
- If passive/passive is configured, neither peer initiates LACP; use the authorized intended active/passive or active/active combination.
- If members suspend, compare speed, duplex, access/trunk/native/allowed settings and L2/L3 role on both ends. Preserve diagnostic messages before changing anything.
- If the routed branch fails, check that physical members are unaddressed and the /30 belongs to the logical channel. Stop on unsupported virtual features.
- If one-member failure drops the whole aggregate, inspect actual remaining participation and any minimum-links policy. Do not force a predicted successful result.
Use an existing Cisco Modeling Labs 2.10 entitlement and its bundled reference images inside CML only. GNS3 does not provide Cisco images; use images there only when a separate applicable license permits that environment. Do not share or redistribute image files. A CML subscription alone does not authorize exporting its Cisco images to GNS3. Without entitlement, use the explicitly labeled paper exercises; they do not establish hands-on competence.
- Restore Gi0/0 on SW1 and RSW1 and verify both members in both independent pairs before stopping nodes.
- Keep actual state and tests in private notes; do not leave shutdown faults saved or export the Cisco reference images.
Practice set
Answer before you reveal
01Can two passive LACP peers initiate a bundle?
At least one active side is needed, while passive can forward normally after negotiation with an active peer.
02Where is a Layer 3 aggregate's address configured?
The members provide transport for one logical IP attachment and must have compatible routed behavior.
03Does one working ping prove both links are bundled?
Inspect summary flags and partner state to establish the intended number of participants.
04Why can one member show more traffic?
A small number of flows need not distribute evenly, and one flow does not automatically combine all member rates.
05What must match for a trunk bundle?
Consistency prevents different members of one logical interface from providing conflicting VLAN services.
06Why not repair suspension with mode on?
Forcing aggregation can introduce loss or loops and is not a defensible substitute for diagnosing the mismatch.
Knowledge check
Quiz: prove the reasoning
Keep these
Technical takeaways
- LACP negotiation requires at least one active initiator.
- Member compatibility matters more than the local channel number.
- Layer 2 carries VLANs; Layer 3 puts addressing on the aggregate.
- Verify every intended member, not just logical ping success.
- A single flow does not automatically consume summed member bandwidth.
- Restore a failed member and verify negotiation before claiming full recovery.