Can't connect to COORDINATOR_HOST when using emergency role token

[Andrew_Riley]

I'm trying to recover my ECE instance after losing my co-ordinator server.

The information on Using Emergency Roles token says to run

install --coordinator-host 192.168.50.10 --roles-token {token}

specifying the old co-ordinator host, but the installation validation output says;

Checking coordinator connectivity ... FAILED
Can't connect COORDINATOR_HOST [ip.address] connection refused

I don't understand this message, because losing the coordinator host is why I'm using the emergency all roles token. What am I misunderstanding?

[Alex_Piggott]

Hmm the docs aren't worded great here ... although it says "COORDINATOR_HOST", in practice it means "an ECE host with the director role".

If you lose 100% of your nodes running a director, then the ECE is gone forever unfortunately .. that's the reason step 2 after install is making ECE HA by adding additional directors

(The main purpose for the emergency token in practice is if you lose all your coordinators - since then the system is happy except that you've lost API access, ie the ability to create tokens)

If you have other director nodes still alive then use their IPs instead

(if you have 3 directors and have lost 2 of them, then the situation can be rectified but it requires a bit of hand editing of files)