[Bug] IpAuthHandler blocks all cross-node raft connections in Docker bridge mode — hostname vs IP mismatch

[bitflicker64]

Bug Type (问题类型)

None

Before submit

• 我已经确认现有的 Issues 与 FAQ 中没有相同 / 重复问题 (I have confirmed and searched that there are no similar problems in the historical issue and documents)

Environment (环境信息)

Environment

• HugeGraph PD version: 1.7.0
• Docker network mode: bridge (static IPs via ipam)
• Tested on: macOS Apple M4
• related issue - [Bug] Docker setup broken on macOS/Windows — network_mode: host replaced with bridge networking (single-node + 3-node cluster) #2951
• Related PR: refactor(docker): migrate single-node compose from host to bridge networking #2952
• Related bug: IpAuthHandler hostname vs IP (same bridge mode root cause)

Expected & Actual behavior (期望与实际表现)

Expected: Cross-node raft connections between PD nodes work correctly in
Docker bridge network mode.

Actual: All cross-node raft connections are silently blocked by
IpAuthHandler. The allowlist is built using PeerId::getIp which returns
raw hostnames (e.g. "pd0", "pd1") but incoming connections arrive with
actual bridge IPs (e.g. "172.18.0.4") — the string comparison always
fails so every connection is dropped with no useful error.

Error in logs:

WARN IpAuthHandler - Blocked connection from 172.18.0.4

Workaround: assign static IPs to PD containers via docker-compose ipam
and use those IPs directly in HG_PD_RAFT_PEERS_LIST instead of hostnames.