[j-nsp] srx cluster - port channel - cisco switches - esx devices - 12.1x45-d15.5 some virtual machines can't be reached

[Ben Dale]

On 24 Oct 2013, at 5:48 pm, pkc_mls <pkc_mls at yahoo.fr> wrote:

> Le 24/10/2013 00:55, Ben Dale a écrit :
>> Can you confirm that you have two "active" port-channels configured on the Cisco side, one into each SRX? 
> From the docs I found I was assuming a single port-channel could handle all interfaces attached to the same reth on srx.
> 
> For example, if ge-0/0/2, ge-0/0/3, ge-5/0/2, ge-5/0/3 are attached to reth0, can I connect all the ports to the same etherchannel on the stack ?
> 
> 

No, there will be two sub-LAGs formed from the SRX - one on the active node, and one on the standby node.  To downstream devices, these appear as distinct LACP bundles, even though they are part of the same reth interface on the SRX.

There is a rather wordy explanation of it here, but the Note box in the middle of the page is probably a good summary:

http://www.juniper.net/techpubs/software/junos-security/junos-security10.2/junos-security-swconfig-interfaces/understand-lacp-in-cc-mode-section.html

Ben