https://bugzilla.novell.com/show_bug.cgi?id=704666 https://bugzilla.novell.com/show_bug.cgi?id=704666#c12 Marius Tomaschewski <mt@novell.com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |lnussel@novell.com, | |mvidner@novell.com --- Comment #12 from Marius Tomaschewski <mt@novell.com> 2011-08-09 16:55:47 UTC --- When RUN_PARALLEL="yes" in set in /etc/sysconfig/boot (default) it is normal (intentional) that the X11 session already starts before the network is up (still waiting for dhcp lease). You may set RUN_PARALLEL=no. Not normal is such a long delay of > 1 minute. See also bug 697929 -- there is no check / ifup does not wait until IPv6 address becomes usable (still in tentative mode). But it doesn't explain the long delay. Please disable SuSEfirewall2 using following commands: insserv -r /etc/init.d/SuSEfirewall2_init insserv -r /etc/init.d/SuSEfirewall2_setup and reboot. Does it help? (In reply to comment #4)
After reading the man page "man ifcfg" I changed the file /etc/sysconfig/network/ifcfg-eth0 into: BOOTPROTO='static' IPADDR='192.168.1.40/24' IPADDR_1='2001:980:7480:1:192:168:1:40/64' STARTMODE='nfsroot'
After "ifdown eth0" and "ifup eth0" the eth0 does have three global IPv6 addresses; the two in the previous comment and the newly specified one.
Is STARTMODE='nfsroot' intentional or is it just set up by yast2 because of bug 677251? In "nfsroot" mode, IP addresses are never removed (a filesystem or volume is assumed to be accessed using one of them). When you don't need this special feature, switch back to STARTMODE=auto. (In reply to comment #9)
I did a test by configuring nothing in the network, so I got a system with one Ethernet interface, which is set to get all its parameters through DHCP, both v4 and v6. But also DNS and routing come through DNS.
OK.
Getting the IPv4 parameters is almost immediately, just a few seconds during booting, however getting the IPv6 parameters takes sometimes 10 minutes.
Hmm... Which router are you using? Please install "radvd" and dump the config provided by the router using "radvdump eth0".
After searching a lot of IPv6 information I found a commented out line in /etc/dhclient6.conf, which after removing the two hashes, solved the problem of the long delay. The line reads "##send dhcp6.rapid-commit;". It still takes more than getting the parameters for IPv4, but as soon as I can log in, the IPv6 network is ready. So I recommend to have these hashes removed, or maybe conditionally removed.
Yes, for 12.x we can enable rapid commit as default I think -- it should be supported by all dhcp6 servers. (In reply to comment #11)
I reinstalled openSUSE 12.1 M3 and watched the appearance of the Scope:Global IPv6 addresses. At first the appearance was fast. However after a reboot it took about 9 minutes. I removed the two hashes in /etc/dhclient6.conf and rebooted again. There was no apparent change in the time it takes to finally get the Scope:Global IPv6 addresses. The setup is traditional ifup with one Ethernet interface and DHCP4 and 6 enabled in YaST Network Settings. For some reason, I don't understand why, the interface gets two IPv6 addresses. But the removal of the hashes does not (always) work.
If you get one or multiple IPv6 addresses depends on the configuration of the router. Read about "AdvAutonomous" (client allowed to assing address itself) and the "AdvManagedFlag" and "AdvOtherConfigFlag" (-> dhcpv6) in radvd.conf(5) manual page. When you're using DHCPv4 and DHCPv6, the BOOTPROTO should be set to 'dhcp' or to 'dhcp4+dhcp6'. The later one forces to wait until both, dhcp4 and 6 clients finished the IP configurations. In BOOTPROTO='dhcp' mode, success of one of them is sufficient. -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.