Hi,
When e1000g was plumbed, it would show the link "unknown" state unless
it explicitly receives link up or link down (such as plugging out the
cable) interrupts from the NIC. So at a certain point, the NICs on your
machine sent the link down interrupts. Could you try to install the quad
cards on other systems? Does it make a difference?
Miles
Bogdan ?ulibrk wrote:
Hello,
here's the thing. Dell PE1950 with both pcie slots taken. One is PERC 5e, the
second is Intel quad port gigabit card with 82571EB chipset.
Symptoms:
dladm show-ether
LINK PTYPE STATE AUTO SPEED-DUPLEX PAUSE
bnx1 current up yes 1G-f none
e1000g0 current down yes 0M-h bi
e1000g1 current down yes 0M-h bi
e1000g2 current down yes 0M-h bi
e1000g3 current down yes 0M-h bi
bnx0 current up yes 1G-f none
dladm show-link
LINK CLASS MTU STATE OVER
bnx1 phys 1500 up --
e1000g0 phys 1500 down --
e1000g1 phys 1500 down --
e1000g2 phys 1500 down --
e1000g3 phys 1500 down --
bnx0 phys 1500 up --
none of the cards can't be brought to "up" state.
prtconf snip:
pci8086,25f9, instance #4 (driver name: pcie_pci)
pci10b5,8508, instance #15 (driver name: pcie_pci)
pci10b5,8508, instance #16 (driver name: pcie_pci)
pci1374,38, instance #0 (driver name: e1000g)
pci1374,38, instance #1 (driver name: e1000g)
pci10b5,8508, instance #17 (driver name: pcie_pci)
pci1374,38, instance #2 (driver name: e1000g)
pci1374,38, instance #3 (driver name: e1000g)
e1000g0: flags=1000803<UP,BROADCAST,MULTICAST,IPv4> mtu 1500 index 4
inet 10.0.1.1 netmask ffffff00 broadcast 10.0.1.255
ether 0:e0:ed:12:fd:fc
e1000g1: flags=1000803<UP,BROADCAST,MULTICAST,IPv4> mtu 1500 index 5
inet 10.0.1.2 netmask ffffff00 broadcast 10.0.1.255
ether 0:e0:ed:12:fd:fd
e1000g2: flags=1000803<UP,BROADCAST,MULTICAST,IPv4> mtu 1500 index 6
inet 10.0.1.3 netmask ffffff00 broadcast 10.0.1.255
ether 0:e0:ed:12:fd:fe
e1000g3: flags=1000803<UP,BROADCAST,MULTICAST,IPv4> mtu 1500 index 7
inet 10.0.1.4 netmask ffffff00 broadcast 10.0.1.255
ether 0:e0:ed:12:fd:ff
_______________________________________________
networking-discuss mailing list
[email protected]
