This on AMD64 single core Fedora C10 host, OpenSolaris guest running from iscsi disk, V3.0 built from SVN late last week (3.0.51_OSEr21206) to get around iscsi regression.
Trying to do "pkg image_update" from snv111b, it runs for an hour or so (rather slowly) and then the Ethernet bridge stops working (it is configured with a static IP address; the host has hard wired 100MB/s Ethernet). Rebooting the guest seems to reset the bridge so it will run for another hour or so. This is quite reproducible (done it 3 times already).
Under VirtualBox 2.2.4, it was possible to do "pkg image-update" to get to snv111b without this problem, so this appears to be a regression.
Thought I'd check here before submitting a bug.
ethernet bridge fails after 100MBytes or so
-
baf
- Volunteer
- Posts: 829
- Joined: 27. Sep 2008, 06:18
- Primary OS: Mac OS X Leopard
- VBox Version: PUEL
- Guest OSses: linux,xp,win7
- Location: Luleå or Skellefteå, Sweden
Re: ethernet bridge fails after 100MBytes or so
What type of network card to have have selected in VB? If pcnet try changing to one of the intel types.
Some say: "You learn as long as you live".
My way: "You live as long as you learn".
My way: "You live as long as you learn".
-
fam
- Posts: 18
- Joined: 4. Jul 2009, 03:04
- Primary OS: Fedora 10
- VBox Version: OSE Fedora
- Guest OSses: OpenSolaris
Re: ethernet bridge fails after 100MBytes or so
It already is --
Intel PRO/1000 MT Desktop (Bridged adapter, eth0)
Intel PRO/1000 MT Desktop (Bridged adapter, eth0)
-
fam
- Posts: 18
- Joined: 4. Jul 2009, 03:04
- Primary OS: Fedora 10
- VBox Version: OSE Fedora
- Guest OSses: OpenSolaris
Re: ethernet bridge fails after 100MBytes or so
This looks a bit like Ticket #3318 (new defect), although that was reported against 2.2.2 in NAT mode. There are some differences - here the interface freezes completely - no packets in any direction, and didn't seem to be a problem in 2.2.4, but perhaps the upgrade to snv111b from snv111a didn't involve transferring more than 1/2GB of data like the upgrade from snv111b to snv117 does. So is this a new bug or an addition to #3318?
The interface also seems to be rather slow - at 100MByte/hour, around 200KBit/sec; the real cpu doesn't seem to be particularly stressed. Is this normal? I would have expected at least an order of magnitude higher speed than this. From previous image-updates on X86 and SPARC, I don't think the upload site is doing any throttling.
The interface also seems to be rather slow - at 100MByte/hour, around 200KBit/sec; the real cpu doesn't seem to be particularly stressed. Is this normal? I would have expected at least an order of magnitude higher speed than this. From previous image-updates on X86 and SPARC, I don't think the upload site is doing any throttling.