When trying to activate the wired connection on my Thinkpad X220, I'm hitting the following error. NetworkManager[1278]: <info> Activation (em1) Stage 3 of 5 (IP Configure Start) starting DHCPv6 as requested by IPv6 router... NetworkManager[1278]: <info> Activation (em1) Beginning DHCPv6 transaction (timeout in 45 seconds) NetworkManager[1278]: <info> dhclient started with pid 4243 NetworkManager[1278]: <info> Policy set 'em1' (em1) as default for IPv6 routing and DNS. dhclient[4243]: irs_resconf_load failed: 59. dhclient[4243]: Can't initialize context: invalid address format dhclient[4243]: Jan 23 12:08:45 vorpalblade dhclient[4243]: This version of ISC DHCP is based on the release available dhclient[4243]: on ftp.isc.org. Features have been added and other changes dhclient[4243]: have been made to the base software release in order to make dhclient[4243]: it work better with this distribution. dhclient[4243]: Jan 23 12:08:45 vorpalblade dhclient[4243]: Please report for this software via the Red Hat Bugzilla site: dhclient[4243]: http://bugzilla.redhat.com Version-Release number of selected component (if applicable): kernel-3.13.0-1.fc21.x86_64 NetworkManager-0.9.9.0-25.git20140117.fc21.x86_64 dhclient-4.3.0-0.4.b1.fc21.x86_64
So that appears to be dhclient trying to load /etc/resolv.conf: dhcp-4.3.0b1/omapip/isclib.c: result = irs_resconf_load(dhcp_gbl_ctx.mctx, _PATH_RESOLV_CONF, dhcp-4.3.0b1/omapip/isclib.c: log_error("irs_resconf_load failed: %d.", result); which appears to be used for the auto DNS update stuff. At least for NM, this can be turned off with DHCP_SEND_HOSTNAME=no in the ifcfg file. But Luke, it would be interesting to see what your /etc/resolv.conf file contains that causes dhclient to puke?
(In reply to Dan Williams from comment #1) > But Luke, it would be interesting to see what your /etc/resolv.conf file > contains that causes dhclient to puke? $ cat /etc/resolv.conf # Generated by dnssec-trigger 0.11 nameserver 127.0.0.1 $ cat /etc/resolv.conf.tmp # Generated by NetworkManager nameserver 192.168.1.1 nameserver fe80::6a1:51ff:fee7:1be5%em1
Thanks Dan for the investigation. These problems (bug #1052343) should be fixed with bind-libs-lite-9.9.5-0.4.rc2.fc21 Luke, can you update 'bind*' and retest please ?
Closing as I assume this has been fixed with final dhcp-4.3.0 and bind-9.9.5
Actually I still see it. This time it's because the irs_resconf_load() checks [1] the nameservers' it finds in /etc/resolv.conf and it does not like the %<interface> that NetworkManager adds, see comment #2 or bug #1093294. [1] bind/lib/irs/resconf.c:add_server(address_str) struct addrinfo hints, *res; hints.ai_family = AF_UNSPEC; hints.ai_socktype = SOCK_DGRAM; hints.ai_protocol = IPPROTO_UDP; hints.ai_flags = AI_NUMERICHOST; error = getaddrinfo(address_str, "53", &hints, &res); if (error != 0) return (ISC_R_BADADDRESSFORM);
Created attachment 895131 [details] patch I posted this patch upstream.
[ISC-Bugs #35989]
Created attachment 919051 [details] patch
Tomas, I think you can close this ticket as dhcp-4.3.1b1 with a work-around has been in rawhide for some time.
(In reply to Jiri Popelka from comment #9) > Tomas, I think you can close this ticket as dhcp-4.3.1b1 with a work-around > has been in rawhide for some time. Thanks. Closing as NEXTRELEASE as the fix is included in dhcp package available in rawhide.