A broken EtherChannel rarely announces itself.
Most of the time the bundle quietly fails to form, and your traffic falls back on a single link.You are the engineer on duty, and four tickets are waiting in your queue.
For each one, you will read the symptoms, find the cause, apply the fix, and verify it.
Figure 1 – EtherChannel Lab Topology
Every ticket starts from this topology.
Reading the Flags
Ticket 1: users report poor performance between two access switches.
Both uplinks are cabled and up, yet bandwidth feels like a single link.Open the console of SW1 and start with the summary.
SW1# show etherchannel summary Flags: D - down P - bundled in port-channel I - stand-alone s - suspended H - Hot-standby (LACP only) R - Layer3 S - Layer2 U - in use f - failed to allocate aggregator M - not in use, minimum links not met u - unsuitable for bundling w - waiting to be aggregated d - default port A - formed by Auto LAG Number of channel-groups in use: 1 Number of aggregators: 1 Group Port-channel Protocol Ports ------+-------------+-----------+----------------------------------------------- 1 Po1(SD) PAgP Gi0/0(I) Gi0/1(I)Po1 (SD): the port-channel is Layer 2 and down.
Gi0/0 (I) and Gi0/1 (I): each member gave up on the negotiation and now runs as a plain stand-alone interface.
That matches the ticket: a stand-alone port still forwards traffic like an ordinary link, STP blocks the redundant one, and the network keeps working.
Nothing is down, you have only lost the aggregation.Run the same check from SW2.
SW2# show etherchannel summary Flags: D - down P - bundled in port-channel I - stand-alone s - suspended H - Hot-standby (LACP only) R - Layer3 S - Layer2 U - in use f - failed to allocate aggregator M - not in use, minimum links not met u - unsuitable for bundling w - waiting to be aggregated d - default port A - formed by Auto LAG Number of channel-groups in use: 1 Number of aggregators: 1 Group Port-channel Protocol Ports ------+-------------+-----------+----------------------------------------------- 1 Po1(SD) PAgP Gi0/0(I) Gi0/1(I)Identical state on both sides: the failure is symmetrical.

Figure 2 – Po1 down, both members stand-alone
You now know what you are chasing: the links are healthy, the negotiation is not.
Answer the question below
Which flag marks a PAgP port that failed to bundle?
Finding the Mismatch
Compare the channel-group line on both sides of the same link.
SW1# show running-config interface g0/0 Building configuration... Current configuration : 154 bytes ! interface GigabitEthernet0/0 switchport trunk encapsulation dot1q switchport mode trunk channel-group 1 mode auto endSame command on SW2.
SW2# show running-config interface g0/0 Building configuration... Current configuration : 154 bytes ! interface GigabitEthernet0/0 switchport trunk encapsulation dot1q switchport mode trunk channel-group 1 mode auto endBoth sides sit in auto.
Straight from the mode-compatibility table in the previous lesson: auto plus auto is the dead cell.
Each side waits for the other to speak first, and nobody ever does.
Figure 3 – Auto on both sides, nobody initiates
To fix it, one side has to initiate.
Answer the question below
Which PAgP mode waits and never initiates?
Fixing the Mode
Move SW1 to desirable.
The change stays inside PAgP, so you can type the new mode directly over the old one.SW1# configure terminal Enter configuration commands, one per line. End with CNTL/Z. SW1(config)# interface range g0/0-1 SW1(config-if-range)# channel-group 1 mode desirable SW1(config-if-range)# end %LINK-3-UPDOWN: Interface Port-channel1, changed state to up %LINEPROTO-5-UPDOWN: Line protocol on Interface Port-channel1, changed state to upThe syslog already announces the result.
Confirm it with the summary.SW1# show etherchannel summary Flags: D - down P - bundled in port-channel I - stand-alone s - suspended H - Hot-standby (LACP only) R - Layer3 S - Layer2 U - in use f - failed to allocate aggregator M - not in use, minimum links not met u - unsuitable for bundling w - waiting to be aggregated d - default port A - formed by Auto LAG Number of channel-groups in use: 1 Number of aggregators: 1 Group Port-channel Protocol Ports ------+-------------+-----------+----------------------------------------------- 1 Po1(SU) PAgP Gi0/0(P) Gi0/1(P)Confirm that SW2 agrees before you close anything.
SW2# show etherchannel summary Flags: D - down P - bundled in port-channel I - stand-alone s - suspended H - Hot-standby (LACP only) R - Layer3 S - Layer2 U - in use f - failed to allocate aggregator M - not in use, minimum links not met u - unsuitable for bundling w - waiting to be aggregated d - default port A - formed by Auto LAG Number of channel-groups in use: 1 Number of aggregators: 1 Group Port-channel Protocol Ports ------+-------------+-----------+----------------------------------------------- 1 Po1(SU) PAgP Gi0/0(P) Gi0/1(P)SU on the port-channel, P on both members: the healthy signature from the previous lesson is back.
Desirable plus auto is one of that table's green cells.Answer the question below
Does a stand-alone (I) port still forward traffic?
Ticket 2 makes noise: the bundle went down minutes after a colleague reconfigured SW1 in static mode, while SW2 runs LACP active.
In a panic, he removed his channel-group from SW1 before escalating, so the evidence is gone.
Confirm the story in the configuration, then bring the bundle back.
Figure 4 – EtherChannel Lab Topology
The Blind Bundle
The ticket names a culprit, but open it like any other: summary on both switches.
SW1# show etherchannel summary Flags: D - down P - bundled in port-channel I - stand-alone s - suspended H - Hot-standby (LACP only) R - Layer3 S - Layer2 U - in use f - failed to allocate aggregator M - not in use, minimum links not met u - unsuitable for bundling w - waiting to be aggregated d - default port A - formed by Auto LAG Number of channel-groups in use: 0 Number of aggregators: 0 Group Port-channel Protocol Ports ------+-------------+-----------+-----------------------------------------------Now SW2.
SW2# show etherchannel summary Flags: D - down P - bundled in port-channel I - stand-alone s - suspended H - Hot-standby (LACP only) R - Layer3 S - Layer2 U - in use f - failed to allocate aggregator M - not in use, minimum links not met u - unsuitable for bundling w - waiting to be aggregated d - default port A - formed by Auto LAG Number of channel-groups in use: 1 Number of aggregators: 1 Group Port-channel Protocol Ports ------+-------------+-----------+----------------------------------------------- 1 Po1(SD) LACP Gi0/0(s) Gi0/1(s)No channel-group on SW1, two suspended members on SW2.
The summary tells you the state, never the cause.
Before touching anything, read the configuration on both sides of the link.SW1# show running-config interface g0/0 Building configuration... Current configuration : 129 bytes ! interface GigabitEthernet0/0 switchport trunk encapsulation dot1q switchport mode trunk endThe trunk settings are there, the channel-group line is gone: his rollback left g0/0 outside any bundle, exactly as the ticket said.
Same command on SW2.SW2# show running-config interface g0/0 Building configuration... Current configuration : 157 bytes ! interface GigabitEthernet0/0 switchport trunk encapsulation dot1q switchport mode trunk channel-group 1 mode active endThe configuration confirms it: SW2 waits in LACP active on both ports.
Its members sit suspended with the %ETC-5-L3DONTBNDL2 messages you met in the previous lesson, waiting for LACP answers that never come.One more proof before you replay anything: ask SW2 who answers its LACPDUs.
SW2# show lacp neighbor Flags: S - Device is requesting Slow LACPDUs F - Device is requesting Fast LACPDUs A - Device is in Active mode P - Device is in Passive mode Channel group 1 neighbors Port-channel Po1 neighborsThe table under that last line is empty: SW2 keeps sending LACPDUs and nobody ever answers.
Aligning the Protocol
Everything matches his story: a channel-group in mode on never sends or answers LACPDUs, so SW2 suspended its members the moment his change landed.
The fix: recreate the bundle on SW1 with the protocol SW2 already runs.SW1# configure terminal Enter configuration commands, one per line. End with CNTL/Z. SW1(config)# interface range g0/0-1 SW1(config-if-range)# channel-group 1 mode active Creating a port-channel interface Port-channel 1 SW1(config-if-range)# end %LINK-3-UPDOWN: Interface Port-channel1, changed state to up %LINEPROTO-5-UPDOWN: Line protocol on Interface Port-channel1, changed state to upPo1 comes up as soon as the first LACPDUs are answered.
Verify on SW1.SW1# show etherchannel summary Flags: D - down P - bundled in port-channel I - stand-alone s - suspended H - Hot-standby (LACP only) R - Layer3 S - Layer2 U - in use f - failed to allocate aggregator M - not in use, minimum links not met u - unsuitable for bundling w - waiting to be aggregated d - default port A - formed by Auto LAG Number of channel-groups in use: 1 Number of aggregators: 1 Group Port-channel Protocol Ports ------+-------------+-----------+----------------------------------------------- 1 Po1(SU) LACP Gi0/0(P) Gi0/1(P)And close with SW2.
SW2# show etherchannel summary Flags: D - down P - bundled in port-channel I - stand-alone s - suspended H - Hot-standby (LACP only) R - Layer3 S - Layer2 U - in use f - failed to allocate aggregator M - not in use, minimum links not met u - unsuitable for bundling w - waiting to be aggregated d - default port A - formed by Auto LAG Number of channel-groups in use: 1 Number of aggregators: 1 Group Port-channel Protocol Ports ------+-------------+-----------+----------------------------------------------- 1 Po1(SU) LACP Gi0/0(P) Gi0/1(P)SU on the bundle, P on both members: the EtherChannel is back to normal, ticket closed.
Answer the question below
Which channel-group mode bundles without checking the neighbor?
Ticket 3, the next morning: monitoring reports both uplinks between SW1 and SW2 down, and nobody touched the cables.
The change log shows the same colleague tried his static configuration again during the night.
Figure 5 – EtherChannel Lab Topology
Spotting Err-Disabled Ports
Open the ticket the way you opened the others: summary on both switches.
SW1# show etherchannel summary Flags: D - down P - bundled in port-channel I - stand-alone s - suspended H - Hot-standby (LACP only) R - Layer3 S - Layer2 U - in use f - failed to allocate aggregator M - not in use, minimum links not met u - unsuitable for bundling w - waiting to be aggregated d - default port A - formed by Auto LAG Number of channel-groups in use: 1 Number of aggregators: 1 Group Port-channel Protocol Ports ------+-------------+-----------+----------------------------------------------- 1 Po1(SD) - Gi0/0(D) Gi0/1(D)And the same on SW2.
SW2# show etherchannel summary Flags: D - down P - bundled in port-channel I - stand-alone s - suspended H - Hot-standby (LACP only) R - Layer3 S - Layer2 U - in use f - failed to allocate aggregator M - not in use, minimum links not met u - unsuitable for bundling w - waiting to be aggregated d - default port A - formed by Auto LAG Number of channel-groups in use: 1 Number of aggregators: 1 Group Port-channel Protocol Ports ------+-------------+-----------+----------------------------------------------- 1 Po1(SD) LACP Gi0/0(D) Gi0/1(D)Both sides report the bundle down, but a D flag never tells you why.
One command lists every err-disabled port and the reason.
SW1# show interfaces status err-disabled Port Name Status Reason Err-disabled Vlans Gi0/0 err-disabled channel-misconfig Gi0/1 err-disabled channel-misconfigThe Reason column hands you the root cause: channel-misconfig.
You get the same story from show logging | include ERR_DISABLE when the status output is not enough.On SW2, the same command comes back empty.
SW2# show interfaces status err-disabled Port Name Status Reason Err-disabled VlansErr-disabled is local: the sanction lives on the switch that misbehaved.
Answer the question below
Is the err-disabled sanction applied on both switches?
Fix, Then Bounce
Compare both sides before you correct anything.
SW1# show running-config interface g0/0 Building configuration... Current configuration : 152 bytes ! interface GigabitEthernet0/0 switchport trunk encapsulation dot1q switchport mode trunk channel-group 1 mode on endAnd the same on SW2.
SW2# show running-config interface g0/0 Building configuration... Current configuration : 157 bytes ! interface GigabitEthernet0/0 switchport trunk encapsulation dot1q switchport mode trunk channel-group 1 mode active endMode on facing mode active: the mismatch is confirmed, and that is what you fix.
Fix the cause first: align both sides on LACP. The plan, in order:
shut down both members
remove the old static channel-group
recreate group 1 in LACP active
bring the ports back up
SW1# configure terminal Enter configuration commands, one per line. End with CNTL/Z. SW1(config)# interface range g0/0-1 SW1(config-if-range)# shutdown %LINK-5-CHANGED: Interface GigabitEthernet0/0, changed state to administratively down %LINK-5-CHANGED: Interface GigabitEthernet0/1, changed state to administratively down SW1(config-if-range)# no channel-group 1 SW1(config-if-range)# channel-group 1 mode active SW1(config-if-range)# no shutdown %LINK-3-UPDOWN: Interface GigabitEthernet0/0, changed state to up %LINK-3-UPDOWN: Interface GigabitEthernet0/1, changed state to up %LINK-3-UPDOWN: Interface Port-channel1, changed state to up %LINEPROTO-5-UPDOWN: Line protocol on Interface Port-channel1, changed state to up SW1(config-if-range)# endThe ports are back up and the protocol is aligned.
Check the summary.SW1# show etherchannel summary Flags: D - down P - bundled in port-channel I - stand-alone s - suspended H - Hot-standby (LACP only) R - Layer3 S - Layer2 U - in use f - failed to allocate aggregator M - not in use, minimum links not met u - unsuitable for bundling w - waiting to be aggregated d - default port A - formed by Auto LAG Number of channel-groups in use: 1 Number of aggregators: 1 Group Port-channel Protocol Ports ------+-------------+-----------+----------------------------------------------- 1 Po1(SU) LACP Gi0/0(P) Gi0/1(P)Verify SW2 as well.
SW2# show etherchannel summary Flags: D - down P - bundled in port-channel I - stand-alone s - suspended H - Hot-standby (LACP only) R - Layer3 S - Layer2 U - in use f - failed to allocate aggregator M - not in use, minimum links not met u - unsuitable for bundling w - waiting to be aggregated d - default port A - formed by Auto LAG Number of channel-groups in use: 1 Number of aggregators: 1 Group Port-channel Protocol Ports ------+-------------+-----------+----------------------------------------------- 1 Po1(SU) LACP Gi0/0(P) Gi0/1(P)Fixing the configuration alone is never enough.
An err-disabled port stays down until you shut and no shut it.
The exam plants this trap in almost every err-disabled question.Answer the question below
Do err-disabled ports come back on their own once the configuration is fixed?
Why the Guard Fired
The bundle is back, but you still owe the post-mortem: why did SW1 shut its own ports during the night?
His mode on brought Po1 up on SW1 without asking SW2 anything.
One side saw a single logical link, the other side saw two independent ports.
Figure 6 – SW1 bundles blindly, SW2 keeps two suspended ports
SW1 sent its BPDUs over Po1, hashed onto one member like any other traffic.
SW2 answered with one BPDU per physical port, because for SW2 those ports were independent.SW1 received two different BPDUs on what it believed was a single interface.
A logical port answering as two separate ports is the signature of a channel misconfiguration.
Figure 7 – Two BPDUs, one logical interface
Cisco switches watch for that signature with EtherChannel guard, enabled by default.
The moment it fired, the console filled up.%PM-4-ERR_DISABLE: channel-misconfig (STP) error detected on Po1, putting Gi0/0 in err-disable state %PM-4-ERR_DISABLE: channel-misconfig (STP) error detected on Po1, putting Gi0/1 in err-disable state %PM-4-ERR_DISABLE: channel-misconfig (STP) error detected on Po1, putting Po1 in err-disable state %LINK-3-UPDOWN: Interface Port-channel1, changed state to down
Figure 8 – The guard err-disables the static side
The sanction hits the side that bundled blindly, and only that side: that is why SW2 showed no err-disabled port.
It also explains the dash in the Protocol column: mode on negotiates nothing.Answer the question below
How many BPDUs does SW2 send for the two links?
Automating Recovery
Bouncing ports by hand works fine during the day.
At 3 AM, you want the switch to retry on its own.SW1# configure terminal Enter configuration commands, one per line. End with CNTL/Z. SW1(config)# errdisable recovery cause channel-misconfig SW1(config)# endNow check what the recovery engine watches.
SW1# show errdisable recovery ErrDisable Reason Timer Status ----------------- -------------- bpduguard Disabled channel-misconfig (STP) Enabled dtp-flap Disabled link-flap Disabled pagp-flap Disabled psecure-violation Disabled security-violation Disabled Timer interval: 300 secondsEvery 300 seconds, the switch re-enables the err-disabled ports and lets them try again.
Recovery retries, but it never repairs.
If the misconfiguration is still there, the guard fires again and the port goes right back down.Answer the question below
Does errdisable recovery fix the root cause?
Ticket 4 is the sneakiest of your shift.
Po1 is up, every member shows P, and users still complain that transfers crawl.
Figure 9 – EtherChannel Lab Topology
Open the ticket with the usual reflex: summary on both switches.
The legend has nothing left to teach you, so filter straight to the table with begin Group.SW1# show etherchannel summary | begin Group Group Port-channel Protocol Ports ------+-------------+-----------+----------------------------------------------- 1 Po1(SU) LACP Gi0/0(P) Gi0/1(P)Same check on SW2.
SW2# show etherchannel summary | begin Group Group Port-channel Protocol Ports ------+-------------+-----------+----------------------------------------------- 1 Po1(SU) LACP Gi0/0(P) Gi0/1(P)Everything is green on both sides, yet the ticket is real.
One Flow, One Link
The switch picks a member for each frame by running a hash on fields of that frame: MAC addresses, IP addresses, ports.
Same fields in, same member out.A single flow always lands on the same member link.
That is deliberate: spraying one flow across links would reorder packets and wreck TCP.
Figure 10 – Many small flows spread across both members
With varied fields, the hash keeps both members busy.
Answer the question below
Can one flow use two member links at the same time?
When the Hash Can't Help
Now picture the nightly backup: one server, one destination, one massive flow.
The hash computes the same result for every frame and pins the whole flow to one member.One link saturates while the other sits idle: a single flow can't be split across members, however good the hash.

Figure 11 – One massive flow saturates a single member
Check what your switch hashes today.
SW1# show etherchannel load-balance EtherChannel Load-Balancing Configuration: src-mac EtherChannel Load-Balancing Addresses Used Per-Protocol: Non-IP: Source MAC address IPv4: Source MAC address IPv6: Source MAC addressSource MAC only.
Every frame from the backup server hashes identically, whatever the destination.Answer the question below
Which load-balancing method is the switch using before your fix?
Picking a Better Hash
Change the method globally, and pick fields that actually vary in your traffic.
SW1# configure terminal Enter configuration commands, one per line. End with CNTL/Z. SW1(config)# port-channel load-balance src-dst-ip SW1(config)# endNow read the method back.
SW1# show etherchannel load-balance EtherChannel Load-Balancing Configuration: src-dst-ip EtherChannel Load-Balancing Addresses Used Per-Protocol: Non-IP: Source XOR Destination MAC address IPv4: Source XOR Destination IP address IPv6: Source XOR Destination IP addressBefore applying the same fix on SW2, check its method too.
SW2# show etherchannel load-balance EtherChannel Load-Balancing Configuration: src-mac EtherChannel Load-Balancing Addresses Used Per-Protocol: Non-IP: Source MAC address IPv4: Source MAC address IPv6: Source MAC addressSame src-mac on SW2. Align it.
SW2# configure terminal Enter configuration commands, one per line. End with CNTL/Z. SW2(config)# port-channel load-balance src-dst-ip SW2(config)# endThe method is global to the switch and local to each direction.
SW1 hashes the traffic it sends and SW2 hashes its own, which is why you just configured both sides.One proof is still missing. Ask the switch which member it would pick for the backup server, then for a second destination.
SW1# test etherchannel load-balance interface port-channel 1 ip 192.168.50.10 192.168.50.20 Would select Gi0/0 of Po1 SW1# test etherchannel load-balance interface port-channel 1 ip 192.168.50.10 192.168.50.21 Would select Gi0/1 of Po1Same source, two destinations, two different members: the flows can finally spread.
With the old src-mac method, both tests would have returned the same member, whatever the destination.That closes the ticket queue. Whatever the ticket, the reflex never changes: summary on both switches, configuration before correction, and a final check to close.
Answer the question below
Is the load-balancing method global or per port-channel?
Answer the question below
With src-dst-ip, which member would the test pick for 192.168.50.20 to 192.168.50.10?