I have a VM where the internal IP-space created by the NAT is required to be the same across several hosts.
In order to do this, I change the network setup as detailed in the following link:
http://communities.vmware.com/message/2155960#2155960
The precise order of execution is:
- Install VMPlayer (5.0.1 build-894247) on computer (Windows 7 Enterprise, 64-bit 6.1.7601, Service Pack 1)
- Copy the guest vm-files to the computer, set up to use NAT, DHCP and DNS from host
- Start up guest, log in and note that the IP address is different. Access via NAT works (ping, nslookup, http etc)
- Shut down guest
- Change the NAT (VMnet8) settings per link above using the Virtual Network Editor, click Apply && OK
- Start guest, log in and note that the IP address is now changed to the designated. Ping the host internal IP succesfully. Fail to do nslookup, ping (and hence http) to any previously working addresses outside the virtual network. I can access the guest address from the host (eg. http)
- Shut down the guest
- Fire up the Virtual Network Editor, click on Restore Default, wait for a moment
- Start the guest, log in, now the access to any external address works again as in #3
I have also enabled the ip-routing on the host machine, without any difference in behaviour detailed above.
Is there some step that I am missing here?
It also seems that restoring the defaults does much more than what is done when the changed settings are applied from the Virtual Network Editor which is in my mind a bit suspicious.
Poltsi
Update 1.
This got me stymied: I decided to uninstall the VMPlayer from the host, and checked the configuration for removal too.
After the reinstall, I first reran the Virtual Network Editor, modified the settings again, and saved. After that I started a virtual machine, and now it was able to both resolve, ping and access external hosts.
I closed down this guest, and started another one, which then reverted to the nonfunctional behaviour described above.
I closed down the second guest and started the first one. Now this one also has a non-working network.
Could this be a bug in the implementation?