OSX 10.5.4 Guest: Windows XP SP2 Parallels 3 5584 Cisco VPN 3.9 The Cisco VPN connection is NOT a split-tunnel - all of my traffic is going through the VPN. If the VPN is connected, the guest OS does some funky stuff: -DNS WILL work -tracert will return one hop to any destination, ie: Tracing route to google.com [64.233.167.99] over a maximum of 30 hops: 1 89 ms 92 ms 89 ms py-in-f99.google.com [64.233.167.99] Trace complete. -No web pages will display, and traffic seems to get hung while trying to get from guest OS to anywhere. As soon as I disconnect from the VPN, the Internet works fine. I am using 'shared internet connection' (i have tried bridged as well - no luck). I have tried the following to resolve: -Start VPN then start Parallels -Start VPN, Disconnect Windows Connection, Re-Connnect windows connection -Parallels in ALL network modes - with and without internet sharing enabled I feel like what is happening is that there is a new DEV that is created for parallels on the host, somehow when traffic is going out, its not being picked up by the Cisco VPN kernel module... But that doesnt make sense because DNS is working. I have no idea. Has anyone seen this before? Any known fixes? Thanks in advance.
I suppose you need to try split-tunneling, in this case Internet will work from Windows XP, but as you need Internet through VPN, I suppose there are some restriction on VPN (Cisco)server side by MAC address which prevents NAT to work in this case
The problem was not with the VPN (server side) - there is no restrictions for MAC addresses (infact, NATting would replace the VM mac wiht the Host MAC before it gets sent out). Regardless, no internet (private vpn side, or public side) would work. The fix for me was to install the cisco client on the Guest VM. There must be some funky network hooks they use in order to route traffic correctly. I dont even need to connect to the vpn on the guest, just have the client installed...
The problems is with parallels. I am having this same issue. What happens, is the Host system has a DNS of say 1.1.1.1, but when you connect with cisco vpn client, cisco vpn client adapter which your network goes out thru when in vpn, has say dns of 2.2.2.2. Yet parallels guest os still pulls DNS of 1.1.1.1. so the problem is that the parallels guest OS is using the original DNS servers, instead of the new DNS enforced by the vpn adapter. Most vpn adapters work this same way so I would think parallels would have an issue with all of them. I could have sworn I had this working by changing from shared to bridged networking, but it doesn't appear to be the case.
try to use ipconfig /rellease and ipconfig /renew Please note, problem is not with Parallels but in Windows implementation of TCP/IP
I am also seeing the same thing, with the Cisco client, a vpnc client, and the built-in PPTP client in OS X 10.5.5. It's clearly that the VM is not picking up the DNS information properly, because I cannot look up anything in my internal network, but can if I manually specify the internal DNS machines in the TCP/IP properties of the Windows VM. On the OS X side, I can look up all these machines, so the information is clearly getting there. This happens whether I'm using shared or bridged networking in the VM, and doing ipconfig /release and then /renew makes no difference (which it shouldn't, since the VM was started after the VPN connection was made). Best, David
