rdmsm4x-changelog-20260904-2116-claude-fleet-quota-and-ssh-fallback-verified

Enhanced fleet SSH fallback aliases and verified continuous multi-node Claude quota monitoring and active agent execution.

Scope

Hosts: rdmsm4x (canonical coordinator), rdmbair15m5, rdmbair13m5, jdmbair13m5, rdmpw3265m, rdmpw3275m
Subsystems: SSH client configuration, quota monitoring daemon, fleet coordination bus

~/.ssh/config: Aliased bare fleet hostnames (rdmbair15m5, rdmbair13m5, jdmbair13m5, rdmpw3265m, rdmpw3275m) to their respective Tailscale IPs under the LAN split-subnet fallback block, ensuring non-interactive SSH operates consistently regardless of whether .local is passed.
~/scripts/claude_quota_monitor_resume.py: Updated spoke health inspection to use bare hostnames with bounded 5s connection timeouts and 8s overall command timeouts.

ssh -o ConnectTimeout=4 <host> "scutil --get ComputerName" tested across all 5 spokes.
python3 ~/scripts/claude_quota_monitor_resume.py verified across local and all 5 remote nodes.

SSH connectivity verified to 5/5 fleet spokes (rdmbair15m5, rdmbair13m5, jdmbair13m5, rdmpw3265m, rdmpw3275m).
Monitor execution confirmed:
- 5-hour Anthropic session window: 2.0% (resets 02:00 EDT / 06:00 UTC).
- 7-day Anthropic weekly window: 21.0%.
- Local active Claude instances: 42 processes, 51 active/recent sessions.
- Active Claude Desktop orchestrator session 65c5e356 confirmed communicating with Anthropic API over HTTPS without rate limits, actively running repository sweeps and ticket processing.
LaunchAgent com.eastcoastscience.claude-quota-monitor verified loaded and cycling every 300s.

~/.ssh/config: Backup retained at ~/.ssh/config.bak (or revert aliases to .local only).
Scripts located at ~/scripts/claude_quota_monitor_resume.py.