LACP Issue?

[jethro.w.tambeana]

I setup x4 N5K switches, in pairs as shown below.

[N5k-1]------[N5k-2]

|                       |

|                       |

|                       |

|                       |

[N5k-3]------ [N5k-4]

below are the configs for the vpc on the nexus n5k switches. it seem to have worked for a while then recently one of the links went offline and affected servers connected on the n5k switches. the switch port on one of the n5k switch that connects to the other pair of n5k switches (link from n5k1 --> n5k3) has the status of "Suspended (no LACP PDU)". I can't figure out why that link is down. 

any help would be greatly appreciated. 

=========== n5k-1 ===========
vpc domain 10
role priority 2000
system-priority 4000

interface port-channel1
description vPC Peer Link
vpc peer-link

interface port-channel2
description vPC link to second N5k pair
vpc 2

interface eth 1/1
description N5k-3 Eth 1/1
channel-group 2 mode active

=========== n5k-2 ===========
vpc domain 10
role priority 2100
system-priority 4000

interface port-channel1
description vPC Peer Link
vpc peer-link

interface port-channel2
description vPC link to second N5k pair
vpc 2

interface eth 1/2
description N5k-4 Eth 1/2
channel-group 2 mode active

=========== n5k-3 ===========
vpc domain 20
role priority 3000
system-priority 4000

interface port-channel1
description vPC Peer Link
vpc peer-link

interface port-channel2
description vPC link to first N5k pair
vpc 2

interface eth 1/1
description N5k-1 Eth 1/1
channel-group 2 mode active

=========== n5k-4 ===========
vpc domain 20
role priority 3100
system-priority 4000

interface port-channel1
description vPC Peer Link
vpc peer-link

interface port-channel2
description vPC link to first N5k pair
vpc 2

interface eth 1/2
description N5k-2 Eth 1/2
channel-group 2 mode active

[Sergiu.Daniluk]

HIi @jethro.w.tambeana 

You can verify which one is (not) sending/receiving  LACPDUs using:

show lacp counters

Use the command 3-4 times with a span of 60 seconds between each output.

Regards,

Sergiu

[jethro.w.tambeana]

I currently have manually shutdown the port on the nexus which was suspended due to 'no LACP PDUs'. my fear is that turning on the port again to monitor LACP might cause issue with my servers. is it safe to re-enabled the port without causing any issue with our servers connected to the nexus with other port-channels?

[Sergiu.Daniluk]

Also, what exact platform and version are you running?

[jethro.w.tambeana]

all 4 switches are nexus 5596T and 3 are running system version 7.1(3)N1(1) and one is running a older firmware 7.0(2)N1(1)

[Sergiu.Daniluk · Accepted solution]

@jethro.w.tambeana 

Ok, hold on a bit. :-) You have a vPC domain with one peer running 7.1.3 and the other 7.0.2.

This is something which is not supported (other then the short period of time of doing the upgrade).

Before continuing with any troubleshooting steps, you will need to bring all switches to same running version.

Stay safe,

Sergiu

[jethro.w.tambeana]

email noted. i will proceed with the upgrade that one n5k to be the same with the other 4 n5k. 

question:

1. can i extract the firmware version 7.1.3 from one of the 3 n5k switches running the 7.1.3 version and upload it onto that one n5k switch that currently runs 7.0.2 version?

2. Can I have a link for a guide to do the firmware upgrade for the n5k switches please.

Jethro 

[jethro.w.tambeana]

upgrading all switches to same firmware version solved my problem. thank you for your assistance.

[paul driver]

Hello
show feature
conf t
feature lacp

Also try the following
no lacp graceful-convergence - allows for a longer wait period for lacp port initialization

int port x
no lacp suspend-individual - this helps prevent loops through bad switch config -  when disabled the port is prevented from suspending if it doesn’t receive a correct lacp packet- so use with caution!

[Abdulila Alhosaine]

You need to use the same IOS on all the switches