After updating StartOS from 0.4.0 to 0.4.0.2, I lost dashboard access from a Windows PC on another VLAN. The same PC and server IP worked before the update.
Network:
- Windows PC: 10.0.3.100
- StartOS: 10.0.11.2, on the Nodes VLAN (10.0.11.0/28)
- Router: UniFi UCG Max
- Miners on the same VLAN as StartOS continued connecting successfully.
My existing firewall rule allows TCP 80 and 443 from 10.0.3.100 to 10.0.11.2, with automatic return traffic enabled. UniFi recorded hits on this rule, and its gateway ping test received replies from StartOS.
Before the workaround, Windows Test-NetConnection reported TcpTestSucceeded: False on both 80 and 443. Browser access to https://10.0.11.2/ timed out.
Workaround: Adding a UniFi masquerade rule with source 10.0.3.100, destination 10.0.11.2, and outbound interface Nodes restored dashboard access.
StartOS shows its local IP and .local HTTPS dashboard addresses enabled, with public dashboard access disabled.
The release notes mention changes for services accessed across routed private networks/VLANs. Could those changes have affected dashboard traffic or reply routing? Could my PC’s 10.0.3.100 address overlap StartOS’s internal container network?
I haven’t collected server routes or firewall rules, so the cause is unconfirmed. Is there a supported fix that removes the need for masquerade, and what diagnostics would help?