packet loss w/ OS 2009.06 guest on OS X host
-
- Posts: 3
- Joined: 29. Oct 2009, 03:17
- Primary OS: Mac OS X Leopard
- VBox Version: PUEL
- Guest OSses: OpenSolaris, Ubuntu, OpenBSD
packet loss w/ OS 2009.06 guest on OS X host
As of yet, I've not been able to make network connectivity sufficiently reliable to install the emacs package from opensolaris.org. I'm using bridged networking, and I'm seeing packets dropped pinging an IP belonging to the host OS on the same NIC. In contrast, I seem to be able to download the complete source tree for OpenBSD 4.6 from a system running that OS. Ubuntu 9.10 also experiences packet loss, but not at quite such high rates. I'm using the VT-x extensions on all VMs. I've tried both e1000g and pcn drivers with pretty much all of the chipsets (don't think I've tried the earlier PCN chipset as of yet). I've also tried reducing the CPU count from 2 to 1 and not seen any result. I saw this problem on 3.0.8 and continue to see it after upgrading to 3.0.10.
-
- Posts: 1
- Joined: 6. Mar 2010, 03:48
- Primary OS: Mac OS X other
- VBox Version: PUEL
- Guest OSses: Opensolaris
Re: packet loss w/ OS 2009.06 guest on OS X host
I experienced the same issue for all NIC types on my Mac Pro (oddly, I did not experience this issue on my Macbook Air). I solved it by editing /kern/drv/e1000g.conf and changing AutoNegAdvertised, MaxFrameSize, and TbiCompatibiltyEnable to reasonable alternative values. One or more of those is the culprit. Clearly there is a bad interaction at the link layer between OpenSolaris' drivers and OS X's.
Re: packet loss w/ OS 2009.06 guest on OS X host
I too have occasional loss of network functionality when running a OpenSolaris guest (snv_134) in a Virtual Box (3.1.6) VM. Would you mind posting your e1000g.conf settings?kaiser wrote:I experienced the same issue for all NIC types on my Mac Pro (oddly, I did not experience this issue on my Macbook Air). I solved it by editing /kern/drv/e1000g.conf and changing AutoNegAdvertised, MaxFrameSize, and TbiCompatibiltyEnable to reasonable alternative values. One or more of those is the culprit. Clearly there is a bad interaction at the link layer between OpenSolaris' drivers and OS X's.