3850 switch stack going into boot loop after adding new stack member

[Jagdwulfe]

Hello all,

 I just had an issue arise last night when I had a tech onsite add a switch to a 3850 stack running 03.07.03E. The tech had the new switch already powered up when he connected it to the stack. When I sent out the switch because it was a rush I had only set the switch 1 priority 1 command on it to ensure it would not try to become a stack master. The stack rebooted with the new member and went into a rebooted. According to the tech there user ports were showing traffic but the switch was otherwise unreachable and our ISP stated they are getting an incomplete ARP entry but seeing the port up up to their equipment. I then had the tech separate the actual stack master from the stack(After powering them all down.) and had him boot it up. The stack master was behaving the same way as previously mentioned. All user ports were passing traffic and the port to the ISP was up up but incomplete ARP. I am thinking something got corrupted due to the switch being already powered up and added to the stack but I have not found anything via Google on if this is a known bug or not. Any help would be appreciated.

[tschnell]

My TAC case was SR 686738677.   I went from 16.3.7  when switches shipped to 16.6.4a  with a 4 stack switch set. After upgrade completed,  1 of the 4 switches went into boot loop.  Switch 3 kept saying it could not talk to stack and kept rebooting.  I checked the cables and replaced all of them but nothing worked.  I had to put back on 16.3.7 on all 4 switches and get the stack talking again and redo the upgrade to 16.6.4a .  The second time thru the upgrade worked.