NodeID of pump/drainer is not import and keep consistent when import from ansible

[july2993]

Bug Report

Please answer these questions before submitting your issue. Thanks!

1. What did you do?

import a cluster from ansible.
restart pump or drainer.

2. What did you expect to see?
   ./binlogctl -cmd pumps or ./binlogctl -cmd drainers
   things keep the same as use ansible before.

3. What did you see instead?

new instance of pump/drainer with the new NodeID(the one we see from tiup cluster display)

ansible use the {host}:port, and when imported to tiup, we use the {ip:port} and NodeID

4. What version of TiUP are you using (tiup --version)?

[lucklove]

We can get the real id from pump and drainer api.
I think we can transform the tiup-node-id to ansible-node-id via these apis before we take any action on pump and drainer.

[july2993]

should add a NodeID field in spec, and set it when import form ansible

[glkappe]

Regarding this improvement, I hope to ensure that the modification of tiup_node-id is consistent with the official document (also consistent with the deployment of tidb-ansible), It is recommended to continue to use host:port to set unspecified pump_node-id,and avoid secondary learning

https://docs.pingcap.com/zh/tidb/stable/tidb-binlog-configuration-file#node-id

Thanks, please feel free to contact me if you have any questions

[lucklove]

@glkappe We want to change it to ip:port because all instances in TiUP is addressed by ip:port pair, if Drainer not, it breaks the consistency.