While running tests with gPXE I notice a weird issue with the network connection. I am booting XP over HTTP and boot with gPXE for that. Server for the boot image is on the host side running HFS. Host is xp sp3 with workstation 6.5.3.
Now the issue is that the image transfer will hang for about 5 seconds at predictable intervalls, leading to average transfer speed over 1000 Kb/s slower than necessary. The strange part is that this behaviour only occurs when no physical nic is connected to host. The virtual network connection can be host-only or nat, and makes no difference. Connecting a physical nic to host will eliminate this behaviour. Also booting over bridged connection will eliminate behaviour since a physical nic is connected.
There is no difference in using e1000 or pcnet32, regarding the issue.
Have anybody noticed this?
Is there any logic in this?
Joakim
Hi Joakim
can you attach one or more vmware.log where this issue occurs ?
I'd like to look at them
Ulli
___________________________________
VMX-parameters- VMware-liveCD - VM-Sickbay
Ineresting - only difference I see with regards to network is that in case of the "with -physical nic" example you have a USB network adapter plugged in to the host - the Belkin.
In the example with the dropouts you have 3 usb-storage devices
Hmm - don't see a relation - what do you use the Belkin for ?
___________________________________
VMX-parameters- VMware-liveCD - VM-Sickbay
The belkin is the physical nic (usb wifi) that provide internet connection for the host.
The amount of usb storage devices has been tracked against the issue, and has no impact.
In some kind of way, it would make more sense if the issue appeared with the physical nic connected. At least we could conclude that the usb wifi adapter did bad things...
I will see if I can test this on other virtualization software to observe any difference, and possibly conclude with a vmware bug..
Joakim