Bridge Up takes 30s - any remedy?
Hi,
In order to be able to use qemu's bridged networking without taking my network down, I set up my primary network (eth0) to work through a bridge (br0) right from boot, so that later on the virtual machines can just connect to the bridge and talk to the outside.
The downside is that it now takes 30s for the network to acquire an IP from my DHCP server, whereas before, when using only eth0, it took about 3s.
I've collected the messages that bridge prints out in the kernel, and I'm hoping that you find something I can do about it - maybe specify some parameters so there's no need for bridge to go trial and error.
Anyways, here's the log (I didn't know if the RPC messages were related, so I left them in):
Quoted 9 lines
Nov 4 19:29:55 terra kernel: ACPI: PCI Interrupt 0000:02:01.0[A] -> Link [L3CM] -> GSI 11 (level, low) -> IRQ 11 Nov 4 19:29:56 terra kernel: eth0: Setting promiscuous mode. Nov 4 19:29:56 terra kernel: device eth0 entered promiscuous mode Nov 4 19:29:56 terra kernel: br0: port 1(eth0) entering learning state Nov 4 19:30:03 terra kernel: RPC: sendmsg returned error 101 Nov 4 19:30:11 terra kernel: br0: topology change detected, propagating Nov 4 19:30:11 terra kernel: br0: port 1(eth0) entering forwarding state Nov 4 19:30:16 terra kernel: RPC: sendmsg returned error 101Fri Nov 4 19:30:24 CET 2005
The last line is from using the command 'dhcpcd br0 && date >> /var/log/messages', so I knew how long it took (which turned out to be 29s).
Any hints are greatly appreciated.
Cheers,
~Mik
Signature · 1 line
Pech im Glück, Spiel in der Liebe.
Quoted 27 lines · On Fri, 04 Nov 2005 21:08:36 +0100 Michael Bueker <[email protected]> wrote:
Hi,
In order to be able to use qemu's bridged networking without taking my network down, I set up my primary network (eth0) to work through a bridge (br0) right from boot, so that later on the virtual machines can just connect to the bridge and talk to the outside.
The downside is that it now takes 30s for the network to acquire an IP from my DHCP server, whereas before, when using only eth0, it took about 3s.
I've collected the messages that bridge prints out in the kernel, and I'm hoping that you find something I can do about it - maybe specify some parameters so there's no need for bridge to go trial and error.
Anyways, here's the log (I didn't know if the RPC messages were related, so I left them in):
Quoted 9 lines
Nov 4 19:29:55 terra kernel: ACPI: PCI Interrupt 0000:02:01.0[A] -> Link [L3CM] -> GSI 11 (level, low) -> IRQ 11 Nov 4 19:29:56 terra kernel: eth0: Setting promiscuous mode. Nov 4 19:29:56 terra kernel: device eth0 entered promiscuous mode Nov 4 19:29:56 terra kernel: br0: port 1(eth0) entering learning state Nov 4 19:30:03 terra kernel: RPC: sendmsg returned error 101 Nov 4 19:30:11 terra kernel: br0: topology change detected, propagating Nov 4 19:30:11 terra kernel: br0: port 1(eth0) entering forwarding state Nov 4 19:30:16 terra kernel: RPC: sendmsg returned error 101Fri Nov 4 19:30:24 CET 2005
Quoted 7 lines
The last line is from using the command 'dhcpcd br0 && date >>
/var/log/messages', so I knew how long it took (which turned out to be 29s).Any hints are greatly appreciated.
Cheers,
~Mik
Shorten the forward delay.
Signature · 2 lines
Stephen Hemminger <[email protected]> OSDL http://developer.osdl.org/~shemminger