Multiple Lightsail instances in ap-southeast-1 suddenly stopped making outbound connections on standard ports around June 1, 2026. Services affected: Google OAuth login, AWS S3 uploads, Docker image pulls, apt updates.
What works:
- Inbound TCP (SSH, HTTP, HTTPS) — no issues
- DNS resolution — works
- ICMP (ping to gateway) — works
- TCP to AWS metadata endpoint (169.254.169.254:80) — works
- TCP to port 53 outbound — works
What doesn't:
- Outbound TCP on ports 22, 25, 80, 443, 587, 993 — all timeout
- tcpdump confirms TCP SYN packets leave the interface, no SYN-ACK returns
Port test against 1.1.1.1:
Port 22: TIMEOUT
Port 25: TIMEOUT
Port 53: OPEN
Port 80: TIMEOUT
Port 443: TIMEOUT
Port 587: TIMEOUT
Port 993: TIMEOUT
Port 8080: OPEN
Port 8443: OPEN
What we've tried:
- Instance reboot — no change
- Full Stop/Start from Lightsail console — no change
- Checked iptables/nftables — clean, no DROP rules
- Attached a static IP, Stop/Start with new gateway — no change
- Multiple separate instances (dev + production) affected identically
This looks like transparent packet filtering at the Lightsail network edge. The fact that port 53 works but 80/443/22 don't suggests DPI or port-based filtering, not a generic connectivity failure.
Region: ap-southeast-1
Instance type: $20 Lightsail (2GB)
