Red Hat Bugzilla – Bug 230941
F7t2: IPv6 stateless autoconfig broken w/ network install
Last modified: 2007-11-30 17:11:58 EST
Description of problem: IPv6 stateless autoconfig ("Automatic neighbor discovery") doesn't work when setting up TCP/IP for a network install of Fedora 7 test 2 (either i386 or X86_64). It looks like it is trying to do DHCPv6 instead; selecting DHCPv6 gives identical results. There is no DHCP/DHCPv6 or IPv4 connectivity on this interface. Here's the relevant anaconda log: 14:20:35 INFO : need to set up networking 14:20:35 INFO : going to pick interface 14:20:41 INFO : going to do getNetConfig 14:20:41 INFO : eth1 is not a wireless adapter 14:20:55 INFO : DHCP eth1 - bringing link UP - 11043 65 14:20:55 INFO : DHCPv6: NIC: 0x93e5a0 eth1 0x93e3f0 14:21:41 INFO : DHCPv6 eth1 - state: DHC_TIMEDOUT 14:21:41 ERROR : DHCPv6 eth1 - TIMED OUT - failure. (I also tried with eth0 and got the same result) Version-Release number of selected component (if applicable): Fedora 7 test 2 How reproducible: Every time Steps to Reproduce: 1. Start anaconda from CD (boot.iso) 2. Choose HTTP install source 3. Choose network interface 4. At "Configure TCP/IP" dialog, uncheck "Enable IPv4 support" 5. Leave IPv6 and "Automatic neighbor discovery" enabled, and select "OK" Actual results: Box appears that says "Sending request for IP information for eth1". The box disappears after a minute, and the "Configure TCP/IP" dialog is still there. Expected results: Should automatically get network prefix from radvd server
I should also mention that stateless autoconfig ultimately does seem to work, but only if I choose "Manual configuration" and enter an arbitrary address within my /64. See Bug 230949 for more info.
Fixed in rawhide. Thanks. (reassigned back to anaconda since that's where the bug was)
*** Bug 231502 has been marked as a duplicate of this bug. ***
I tried the boot.iso from this morning's build and it's still broken. When I hit OK at the "Configure TCP/IP" dialog, I see a messagebox appear for an instant and then I'm back at the same dialog. This is what appears in anaconda's log when I do that: CRITICAL: dhcp_nic: both DISABLE_DHCPv6 and DISABLE_DHCv4 (sic) set.
Corner case I wasn't trapping in doDhcp() in net.c in loader. If we reach doDhcp() and the user has caused both DHCP clients to be disabled, I return and push you to the manual network configuration screens (if appropriate). Fixed now in rawhide. Thanks for the feedback.