Best way to migrate cross stack etherchannel mode "on" to "active"

[Michael Muenz]

Hey guys,

I've a cross stack etherchannel with mode on between two datacenters and now want to switch to LACP.
Can I just go into the physical interfaces and do a "no channel-group X mode on" -> "channel-group X mode active" or will I have to shut the ports down or completely remove the Port-Channel interface? Can't find any migration document around and don't have test equipment here. :(

Any ideas?

Thanks! :)

[Leo Laohoo]

Can I just go into the physical interfaces and do a "no channel-group X mode on" -> "channel-group X mode active" or will I have to shut the ports down or completely remove the Port-Channel interface? 

Doing this will cause your etherchannel to go down and any traffic traversing the etherchannel will go nuts.  

Just enter "channel-group <NUMBER> mode active" will be less disruptive.

[Reza Sharifi]

Hi,

Yes, you can change it from "on" to channel-group X mode active or passive without shutting down the portchannel interface. There will be short outage while configuring both sides of the portchannel. with the new config.

HTH

[Michael Muenz]

Hi,

I changed it now to active but they now are working in individual state :(

Group  Port-channel  Protocol    Ports
------+-------------+-----------+-----------------------------------------------
2      Po2(SD)         LACP      Gi1/0/1(I)  Gi2/0/3(I)

Same on the other side, but there's Port-Channel 1 (thought the LACP ID's can differ?).

[Leo Laohoo]

See the "(I)"?  This means "Initializing".  

Post the "sh logs".  The logs can provide valuable clues as to why the links are down.  Port-channel numbers are locally insignificant.  

[Michael Muenz]

Hi Leo,

description tells me that I stands for "stand-alone"

lags:  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

Number of channel-groups in use: 4
Number of aggregators:           4

Group  Port-channel  Protocol    Ports
------+-------------+-----------+-----------------------------------------------
2      Po2(SD)         LACP      Gi1/0/1(I)  Gi2/0/3(I)

I turned on "deb lacp" and "deb lacp events" but I only get:

May 14 04:41:18 MET-1DS: LACP: lacp_insert_partner_cd_inhibitor: didn't change sync flag.
May 14 04:41:18 MET-1DS: LACP: lacp_send_lacpdu: (Gi1/0/1) About to send the 110 LACPDU
May 14 04:41:18 MET-1DS: LACP: lacp_write: LACP 124 bytes out Gi1/0/1
May 14 04:41:45 MET-1DS: LACP: lacp_insert_partner_cd_inhibitor: didn't change sync flag.
May 14 04:41:45 MET-1DS: LACP: lacp_send_lacpdu: (Gi1/0/1) About to send the 110 LACPDU
May 14 04:41:45 MET-1DS: LACP: lacp_write: LACP 124 bytes out Gi1/0/1
May 14 04:42:12 MET-1DS: LACP: lacp_insert_partner_cd_inhibitor: didn't change sync flag.
May 14 04:42:12 MET-1DS: LACP: lacp_send_lacpdu: (Gi1/0/1) About to send the 110 LACPDU
May 14 04:42:12 MET-1DS: LACP: lacp_write: LACP 124 bytes out Gi1/0/1

In the cisco docs I read that when a port comes to mode active and the other side is mode on, it will not join the lacp group. In fact when I change from on to active this would happen or not? Don't know how to debug it.

Thanks

Michael

[Leo Laohoo]

Could you please provide the interface configuration for the two physical interfaces and the etherchannel interface?

Post the configurations for both switches.

[Michael Muenz]

Sure, Switch-A:

Switch Ports Model              SW Version            SW Image
------ ----- -----              ----------            ----------
*    1 30    WS-C3750X-24       12.2(55)SE1           C3750E-UNIVERSALK9-M
     2 30    WS-C3750X-24       12.2(55)SE1           C3750E-UNIVERSALK9-M

interface GigabitEthernet1/0/1
 description X
 switchport trunk encapsulation dot1q
 switchport trunk allowed vlan 18,60,64-66,68,100-102,104,105,200,251,900,902
 switchport trunk allowed vlan add 903,906,911,934,937,939
 switchport mode trunk
 channel-group 2 mode active
end

interface GigabitEthernet2/0/3
 description X
 switchport trunk encapsulation dot1q
 switchport trunk allowed vlan 18,60,64-66,68,100-102,104,105,200,251,900,902
 switchport trunk allowed vlan add 903,906,911,934,937,939
 switchport mode trunk
 channel-group 2 mode active
end

interface Port-channel2
 description X
 switchport trunk encapsulation dot1q
 switchport trunk allowed vlan 18,60,64-66,68,100-102,104,105,200,251,900,902
 switchport trunk allowed vlan add 903,906,911,934,937,939
 switchport mode trunk
end

Switch-B:

*    1 30    WS-C3750X-24       15.0(1)SE2            C3750E-UNIVERSALK9-M
     2 30    WS-C3750X-24       15.0(1)SE2            C3750E-UNIVERSALK9-M

interface GigabitEthernet1/0/1
 description Y
 switchport trunk encapsulation dot1q
 switchport trunk allowed vlan 18,60,64-66,68,100-102,104,105,200,251,900,902
 switchport trunk allowed vlan add 903,906,911,934,937,939
 switchport mode trunk
 channel-group 1 mode active
end

interface GigabitEthernet2/0/1
 description Y
 switchport trunk encapsulation dot1q
 switchport trunk allowed vlan 18,60,64-66,68,100-102,104,105,200,251,900,902
 switchport trunk allowed vlan add 903,906,911,934,937,939
 switchport mode trunk
 channel-group 1 mode active
end

interface Port-channel1
 description Y
 switchport trunk encapsulation dot1q
 switchport trunk allowed vlan 18,60,64-66,68,100-102,104,105,200,251,900,902
 switchport trunk allowed vlan add 903,906,911,934,937,939
 switchport mode trunk
end

[Leo Laohoo]

Everything looks OK.  Hmmmm ... I got a strange feeling I've seen this issue before.  

On both ends, remove the etherchannel config.  SO what you do is this:

config t
 no interface Po <NUMBER>
 interface range <PORT> , <PORT>
 no channel-group <NUMBER> mode active 
 channel-group <NUMBER> mode active

[Michael Muenz]

Strange, isn't it? Was to lazy to read all the caveats for the different releases :)

Next maintenance window is on tuesday, will try to reset it with the no statements.

Thanks

[Leo Laohoo]

Strange because I only see this happening on old IOS for 6500.  Never seen a behaviour like this on a 3750.

[Michael Muenz]

Sorry for coming back so lately, today was next maintenance window.

I dialed via ISDN to console servers and did on both ends:

conf t
int port-ch X
shut
exit
no int port-ch X

(phys. If was shut and no channel config)

Then go into phys If and did a channel-group X mode active, back to port-ch X and added the trunking stuff.

Last a no shut and the ... same thing! Inactive state just STP blocking the vlans. 

This must be a IOS bug, I'll now walk trough the changelog.

[Peter Paluch]

Hi everyone,

@ciscomax: This is a blind shot but can you try to add the VLAN 1 into the list of allowed VLANs on all physical ports that you want to bundle into an EtherChannel on both switches? There is a very remote possibility that because of some IOS issue, LACPDUs (which are sent in the native VLAN and that seems to be VLAN 1 in your config) cannot be sent or received on trunks with the VLAN 1 pruned. This shouldn't happen in normal case but obviously, we're not dealing with normal behavior here.

Best regards,
Peter

[Michael Muenz]

I checked every changelog for LACP or etherchannel, nothing found.
Your shot sounds good, I'll give it a try next maintenance window.

Thanks!