A user brought to my attention that he has the following configuration:
Version: 2.8.6-8602-g.07cdffcaa-0ubuntu1~18.04.1
server01 - region controller + rack controller + vip (keepalived)
server02 - region controller + rack controller + (keepalived)
db - external db
DHCP relay is used
- When server01 is down, deployment fails because "no rack controller can access BMC".
- Rack controller on server02 can access bmc address without problem.
- He manually overwrote /usr/lib/python3/dist-packages/maasserver/models/node.py with hard coded client_idents and fallback_idents, and deployment is able to complete.
It looks like `_get_bmc_client_connection_info` function fails to get appropriate rack controller that can control bmc.
Eric, can you reproduce this bug? If so, can you show me your reproduction of it, and if it seems correct, I'll triage this one? Moving to incomplete in the meantime.