Control: found -1 2.0.4-2 Control: Severity -1 important On 02/01/14 22:26, Alexander Inyukhin wrote: > Package: klibc-utils > Version: 2.0.2-1 > Severity: normal > > Hi! > > I'm trying to boot a machine with two e1000e interfaces and nfsroot. > It works fine with kernel 3.2.0-4-amd64, but fails to get DHCP address > with 3.11.10-1~bpo70+1-amd64. > > Network is attached only to eth1, eth0 is left unconnected. > There is also an IPMI console session over connected interface. > > Manual network configuration from initramfs shell (root is set to > non-existent device to achieve this shell) gives this: > > (initramfs) ipconfig -t 2 -c dhcp -d "" > IP-Config: eth1 hardware address f4:6d:04:03:28:40 mtu 1500 DHCP > IP-Config: eth0 hardware address f4:6d:04:03:27:99 mtu 1500 DHCP > IP-Config: no response after 2 secs - giving up > (initramfs) ipconfig -t 2 -c dhcp -d eth1 > IP-Config: eth1 hardware address f4:6d:04:03:28:40 mtu 1500 DHCP > IP-Config: eth1 complete (dhcp from 192.168.0.254): > address: 192.168.0.18 broadcast: 192.168.0.255 netmask: 255.255.255.0 > > gateway: 192.168.0.254 dns0 : 192.168.0.254 dns1 : 0.0.0.0 > > host : node-18 > rootserver: 192.168.0.254 rootpath: > filename : > > In the first case and "in the wild" ipconfig sends DISCOVER request, but > ignores OFFER. > > To workaround this issue I have to set "IPAPPEND 2" in pxelinux config file. > > >
I have been bitten also by this bug on a fresh Debian Jessie installation while trying to configure the machine to get DHCP in the initramfs. The machine also had two Ethernet interfaces. eth0 connected to the LAN and eth1 with the cable unplugged. After many tries and debugging I ended configuring the IP manually via the boot parameter ip= which was an option for my use case. I'm raising the severity of this because this breaks automatic IP configuration for any machine with more than one internet interface, which is not uncommon nowadays.
signature.asc
Description: OpenPGP digital signature
