SG500 stack LACP failover

[tsi999]

Hello,

We have a pair of SG500 stackable switches to support our cluster. The servers are using LACP etherchannel with 2NICS paired for each channel, one NIC being plugged into master switch and the secondary NIC in the slave to provide redundancy.

When I unplug any single port from a LACP channel, there's no connectivity loss since the secondary port is working and that's as expected.

But, if I simulate a switch fault and reboot either switch, I get total connectivity loss until the switch in question is fully rebooted again.

So, is this by design or am I missing something? I was under impression that the LACP ports on the first switch should still work even if the second switch should totally fail?

[balaji.bandi]

This seems to be configuration issue as Master i Stack, if master fails other switch become master and start forwarding the traffic.

here is example to see configuration :

https://www.cisco.com/c/en/us/support/docs/smb/switches/cisco-small-business-500-series-stackable-managed-switches/smb2553-stack-settings-on-sx500-series-stackable-switches.html

[tsi999]

Thanks, but should this also happen if I reboot the slave?

EDIT: I have rechecked per provided link, the switches use native stacking, L2 mode, with auto master election.

[balaji.bandi]

As per your discussion you have mentioned. when you shutdown Primary port towards server, the secondary port activated.

This proves and assume server side failover taking place, now when you reboot the switch 1, it fail over fails.

can we have more information when failure other switch become master ?

[tsi999]

The LACP works fine when both units are online (I can disconnect any given port without network outages).

But, If I reboot the slave, all cross-stack LACP network teams go down until slave comes up again.

Same if I reboot the master. The stack is auto-electing master, but that shouldn't matter if I reboot the slave in the first place, right?

[balaji.bandi]

Can you provide the configuration of the switch and logs while this take place.

before reboot post show etherchannel X  summary

after one reboot same output also , individual interface status (up or down ?)

[tsi999]

I'm not in position to reboot stuff right now, so I'm providing you with the switch config and the log of rebooting the master from a few days ago. Hope it will suffice.

[balaji.bandi]

Let me see what i can do best with the information provided, what is the Port-channel number connected to Server ?

[tsi999]

All Port channels are connected to different servers. All of them went down during the reboot (I think it's readable from logs that Po # went down.