This message was deleted.
# _lab
s
This message was deleted.
u
tjkreidl: I compiled a new driver based on latest + DDK but, unfortunately, it is just as flakey
u
modinfo asix filename: /lib/modules/3.10.0+10/kernel/drivers/net/usb/asix.ko license: GPL description: ASIX AX88179_178A based USB 2.0/3.0 Gigabit Ethernet Devices author: David Hollis srcversion: 8C067E0D43E821F4DE1F17B
u
USB passtrhough on XenServer was meant more for things like cameras and pointing devices, not for high-throughput devices like NICs. USB kind of works for thumb drives, but that's about as far as you can currently push things.
u
Not looking at passthrough
u
How's the NIC connected then, just to the USB port directly using the DDK driver?
u
USB to Ethernet NIC. XenServer sees it as eth1
u
The VMs see the normal XS interface
u
OK. USB 2 or 3?
u
3
u
Hmmm... not sure why performance should be so bad. Do you see dropouts and such if you run "netstat -s"?
u
BIOS device: eth1 Kernel name: eth1 Permanent MAC: 00249B1EC3:45 Assigned MAC : 00249B1EC3:45 Driver: ax88179_178a Driver version: Firmware version: Bus Info: 2-4:1.0
u
It is worse that that
u
From a windows vm
u
Ping, trace, etc. Everythign works fine
u
Browsing to anything google works (gmail.com)
u
Can telnet to anything over 80
u
Can't browse to anything via IE or chrome, except google
u
Swith back to onboard interface while VM is still up. Everything works
u
I'm currently working to get traces at various locations
s
Lower the mtu?
u
I tested a ping with a packet size of 1472 and it works fine
u
Google.com, gmail works as well. It makes no sense..
s
Google uses IPv6 in a lot of their stuff I think
Any variables to test there?
Can you ping with MTU 1500?
Wire shark?
u
You won't be able to get 1500 mtu
u
because of header size
u
Selgan - I disabled ipv6 and tested previously. But that isn't a bad thought
u
I will be grabbing traces from VM, Host, Router as soon as I have a minute to do so. My guess is that XS is filtering something