rdmbair15m5-changelog-20260915-2122-ssh-fleet-tailscale-fallback

Date / Time: 2026-09-15 21:22:00 EDT
Host: rdmbair15m5 (macOS 27.0, arm64)
Agent: agy@rdmbair15m5
Summary: Aligned ~/.ssh/config on rdmbair15m5 with canonical hub rdmsm4x by adding the fleet .local -> Tailscale fallback blocks, resolving jdmbair13m5 reachability and achieving 100% clean fleet-wide agent_msg.zsh sync.

Rationale

During agent_msg.zsh sync, jdmbair13m5 was reported unreachable because agent_msg.zsh connects to richh@<host>.local. On jdmbair13m5 (currently roaming on Tailscale / Starlink), multicast DNS (.local) does not resolve across the WAN network boundary (ping: cannot resolve jdmbair13m5.local: Unknown host). Canonical hub rdmsm4x had solved this on 2026-09-01 via an explicit Host <name> <name>.local -> HostName <tailscale-ip> fallback block in ~/.ssh/config, but this configuration had not been propagated to rdmbair15m5.

Changes Applied

Backed up existing ~/.ssh/config to ~/.ssh/config.bak-20260915-212147.
Appended the fleet .local -> Tailscale fallback and Non-Interactive Guard blocks to ~/.ssh/config on rdmbair15m5:
- rdmsm4x rdmsm4x.local -> 100.87.66.12
- rdmbair13m5 rdmbair13m5.local -> 100.75.186.59
- jdmbair13m5 jdmbair13m5.local -> 100.86.185.90
- rdmpw3265m rdmpw3265m.local -> 100.113.170.117
- rdmpw3275m rdmpw3275m.local -> 100.95.173.70
- BatchMode yes, ConnectTimeout 5 on fleet hosts.
Synchronized coordination check-ins across the fleet via fleet_checkin_sync.zsh on rdmsm4x.

Verification Evidence

ssh [email protected] 'scutil --get ComputerName' -> jdmbair13m5 (exit 0).
~/.agent-coordination/agent_msg.zsh sync:
- sync: rdmsm4x ok
- sync: rdmbair13m5 ok
- sync: rdmpw3265m ok
- sync: rdmpw3275m ok
- sync: jdmbair13m5 ok (exit code 0 across 5/5 reachable peer hosts).

Rollback / Undo

Delete the marked block between # === BEGIN fleet .local -> Tailscale fallback and # === END Fleet Non-Interactive Guard === from ~/.ssh/config, or restore from ~/.ssh/config.bak-20260915-212147.