vbox + bce == sporactic ethernet hangs

Marc G. Fournier scrappy at hub.org
Tue Apr 23 19:40:11 UTC 2013


I am running FreeBSD 9-STABLE (updated yesterday: FreeBSD 9.1-STABLE #15: Mon Apr 22 07:45:07 UTC 2013) with VirtualBox 4.2.6 from ports … the hardware is using a Broadcom ethernet:

bce0: <Broadcom NetXtreme II BCM5708 1000Base-T (B2)> mem 0xf4000000-0xf5ffffff irq 16 at device 0.0 on pci7
miibus0: <MII bus> on bce0
bce0: Ethernet address: 00:22:19:5b:20:bd
bce0: ASIC (0x57081020); Rev (B2); Bus (PCI-X, 64-bit, 133MHz); B/C (4.4.1); Bufs (RX:2;TX:2;PG:8); Flags (SPLT|MSI|MFW); MFW (UMP 1.1.9)
bce0: bce_pulse(): Warning: bootcode thinks driver is absent! (bc_state = 0x00004006)

Running with simple jail's on it, the server runs flawlessly until reboot … but as soon as I start running Virtualbox on it, I get sporadic server 'hangs' … never the same time, usually can be triggered by heavier then normal load on the virtual box (ie. running an rsync session from the base server into the vbox environment) … 

When it happens, I can *usually* connect via the DRAC / remote console and login … but doing an 'ifconfig down' on the device and then back up makes no difference … if I send a ctl-alt-del through the remote console, more often then not, it will free up whatever is going on, so that pinging works again, but, of course, I've already hit ctl-alt-del, so its rebooting even though now I don't need it to …

Based on a page on the wiki about tuning for vbox, I have set:

net.graph.maxdata=65536

but I've seen this happen even with that set, so not sure if I'm just still triggering it, or its something else I'm experiencing …

So, two questions:

1. is there something I can run to see if I *am* in fact hitting that limit?  
2. is there something I can do, like ctl-alt-del, but without the reboot, to 'free' the ethernet?

Thx




More information about the freebsd-virtualization mailing list