Whilst trying to test a system under high load (100% of 10GE), the following messages were printed out on the console: NOTICE: ixgbe0: ixgbe_rx_copy: allocate buffer failed NOTICE: ixgbe0: ixgbe_rx_copy: allocate buffer failed NOTICE: e1000g1 unregistered NOTICE: e1000g2 unregistered NOTICE: e1000g3 unregistered NOTICE: nxge0 unregistered (it then panic''d) If we ignore the ixgbe messages, what would cause the system to start unregistering the MAC drivers? None of the interfaces being unregistered above were plumb''d, which makes me slightly more curious. Darren
Garrett D''Amore
2010-May-27 15:55 UTC
[crossbow-discuss] What would cause this behaviour?
On 5/27/2010 4:48 AM, Darren Reed wrote:> Whilst trying to test a system under high load (100% of 10GE), > the following messages were printed out on the console: > > NOTICE: ixgbe0: ixgbe_rx_copy: allocate buffer failed > NOTICE: ixgbe0: ixgbe_rx_copy: allocate buffer failed > NOTICE: e1000g1 unregistered > NOTICE: e1000g2 unregistered > NOTICE: e1000g3 unregistered > NOTICE: nxge0 unregistered > > (it then panic''d) > > If we ignore the ixgbe messages, what would cause the > system to start unregistering the MAC drivers? None of > the interfaces being unregistered above were plumb''d, > which makes me slightly more curious.If the system was using a debug kernel, then wild card modunload (-i 0) would cause any attached drivers to unload. Furthermore, some actions (adding a new driver, for example) cause devfsadm to run and attach drivers for all hardware in the system. So, the messages about drivers unregistering should not be considered symptomatic of any problem. -- Garrett> > Darren > > _______________________________________________ > crossbow-discuss mailing list > crossbow-discuss at opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/crossbow-discuss