Customer made stack enviroment using two S6700 switches. Because there is change of network, customer take stack switchover operation. Unfortunately, there is service impacted. They think the switch-setup was designed for failover, but it doesn´t seem to be working.
Collect the logs to analyze the operation process.
a. Customer power off Master switch at 21：54. The Slave switch will become the new Master.
Jul 7 2014 21:54:00+02:00 6700 %%01SRM/4/STACK_PORT_DOWN(l):Stack port STACK 2 change to DOWN.
Jul 7 2014 21:53:59+02:00 6700 %%01SRM/4/STACK_PORT_DOWN(l):Stack port STACK 1 change to DOWN.
b. Power on the old Master switch between 22:04 and 22:07. It finished the reboot at 22:12
c. Power off Slot 0 (Old master). Slot 1(old slave) will change to new master. For stack environment, there is unique SYSTEM MAC.
If Master switch is down and did not recover in 10 minutes, SYSTEM MAC will be changed to new one.
From 21:54 to 22:04, old master is down and did not join the stack system again. SYSTEM MAC is changed.
#Jul 7 2014 22:04:00+02:00 6700 FSP/4/STACK_TRAP:OID 188.8.131.52.4.1.2011.5.25.183.1.22.8 Stack system's Mac address change to stackid 1's Mac address:04f9-3890-fe70.
4. After system mac is changed, system-id will also be changed in LACP packet of eth-trunk.
Peer device received this kind of LACP packet and informed switch to change the status of eth-trunk member ports to unselect. It will make eth-trunk port down. Service is impacted.
After several minutes, peer device informed switch and eth-trunk port status is changed to select. Eth-trunk was up and service is recovered.
In one word, SYSTEM MAC change caused eth-trunk down and impacted the service.