Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!

Bug 39157

Summary: BIND 9.2.2-r2 will not answer on IPv6 interfaces
Product: Gentoo Linux Reporter: Eric Radman <theman>
Component: [OLD] ServerAssignee: Stewart (RETIRED) <blkdeath>
Status: RESOLVED UPSTREAM    
Severity: normal    
Priority: High    
Version: 1.4_rc4   
Hardware: Sparc   
OS: Linux   
URL: http://[2001:470:1f01:ffff::1df]
Whiteboard:
Package list:
Runtime testing required: ---
Attachments: named config file that I'm using
emerge info

Description Eric Radman 2004-01-23 06:44:49 UTC
BIND will lookup AAAA and A6 records, but it will not respond to queries made on an IPv6 socket:

$ dig @127.0.0.1 ipv6.he.net aaaa +short
3ffe:81d0:ffff::3

$ dig @::1 ipv6.he.net aaaa +short
; <<>> DiG 9.2.2 <<>> @::1 ipv6.he.net aaaa +short
;; global options:  printcmd
;; connection timed out; no servers could be reached

The naemon is listening on IPv6 ports:

# netstat -A inet6 -alpn | grep named
tcp        0   0 :::53      :::*          LISTEN      24095/named         
tcp        0   0 ::1:953    :::*          LISTEN      24095/named         
udp        0   0 :::53      :::*                      24095/named         
udp        0   0 :::32872   :::*                      24095/named  

I recompiled bind and bind-tools, and upgraded the kernel to 2.4.24 with no effect.
Comment 1 Stewart (RETIRED) gentoo-dev 2004-01-29 12:51:46 UTC
Not having any IPv6 machines on which to test this; have you queried the bind9-users mailing list at ISC to check for (possible) configuration issues, or an upstream bug? Is this a fresh install, or an upgrade from a prior version of BIND? If the latter; was the prior version working correctly? Further, have you tried un/re-installing BIND?

Meanwhile; can you attach the output of 'emerge info' and perhaps your named.conf? 
Comment 2 Eric Radman 2004-01-30 14:09:25 UTC
When I restart named this error is repeated several times:

Jan 30 14:16:53 stellar named[12121]: internal_send: ::1#36078: Invalid argument
Comment 3 Eric Radman 2004-01-30 14:12:30 UTC
Created attachment 24662 [details]
named config file that I'm using
Comment 4 Eric Radman 2004-01-30 16:34:50 UTC
Created attachment 24666 [details]
emerge info

This was an upgrade, but this is the first time using IPv6 DNS resolution. I
tried re-emering it with no change.
Comment 5 Eric Radman 2004-01-30 19:26:16 UTC
# lsof -i 6 -P -n
COMMAND     PID     USER   FD   TYPE DEVICE SIZE NODE NAME
sshd       1140     root    3u  IPv6   2579       TCP *:22 (LISTEN)
apache2    1150     root    3u  IPv6   2593       TCP *:80 (LISTEN)
apache2    1190   apache    3u  IPv6   2593       TCP *:80 (LISTEN)
apache2    1233   apache    3u  IPv6   2593       TCP *:80 (LISTEN)
couriertc  1239     root    5u  IPv6   2676       TCP *:143 (LISTEN)
couriertc  1288     root    5u  IPv6   2719       TCP *:110 (LISTEN)
master     1625     root   11u  IPv6   3261       TCP *:25 (LISTEN)
pure-ftpd  1671     root    4u  IPv6   3489       TCP *:21 (LISTEN)
rsync      1739     root    4u  IPv6   3661       TCP *:873 (LISTEN)
sshd       1934     root    4u  IPv6   4490       TCP [2001:470:1f01:ffff::1df]:22->[2001:470:1f01:339::211]:35184 (ESTABLISHED)
sshd       1936 sysadmin    4u  IPv6   4490       TCP [2001:470:1f01:ffff::1df]:22->[2001:470:1f01:339::211]:35184 (ESTABLISHED)
sshd       1960     root    4u  IPv6   4696       TCP [2001:470:1f01:ffff::1df]:22->[2001:470:1f01:339::211]:35185 (ESTABLISHED)
sshd       1962 sysadmin    4u  IPv6   4696       TCP [2001:470:1f01:ffff::1df]:22->[2001:470:1f01:339::211]:35185 (ESTABLISHED)
apache2   10988   apache    3u  IPv6   2593       TCP *:80 (LISTEN)
named     13065    named    8u  IPv6  50586       UDP *:53 
named     13065    named    9u  IPv6  50587       TCP *:53 (LISTEN)
named     13065    named   12u  IPv6  50596       TCP [::1]:953 (LISTEN)
named     13065    named   15u  IPv6  50594       UDP *:33030 
named     13070    named    8u  IPv6  50586       UDP *:53 
named     13070    named    9u  IPv6  50587       TCP *:53 (LISTEN)
named     13070    named   12u  IPv6  50596       TCP [::1]:953 (LISTEN)
named     13070    named   15u  IPv6  50594       UDP *:33030 
named     13071    named    8u  IPv6  50586       UDP *:53 
named     13071    named    9u  IPv6  50587       TCP *:53 (LISTEN)
named     13071    named   12u  IPv6  50596       TCP [::1]:953 (LISTEN)
named     13071    named   15u  IPv6  50594       UDP *:33030 
named     13073    named    8u  IPv6  50586       UDP *:53 
named     13073    named    9u  IPv6  50587       TCP *:53 (LISTEN)
named     13073    named   12u  IPv6  50596       TCP [::1]:953 (LISTEN)
named     13073    named   15u  IPv6  50594       UDP *:33030 
named     13074    named    8u  IPv6  50586       UDP *:53 
named     13074    named    9u  IPv6  50587       TCP *:53 (LISTEN)
named     13074    named   12u  IPv6  50596       TCP [::1]:953 (LISTEN)
named     13074    named   15u  IPv6  50594       UDP *:33030 
apache2   19410   apache    3u  IPv6   2593       TCP *:80 (LISTEN)
apache2   19411   apache    3u  IPv6   2593       TCP *:80 (LISTEN)
Comment 6 Eric Radman 2004-01-30 19:48:11 UTC
It seems that something in the IPv6 UDP socket code is broken here:

Jan 30 22:36:33 stellar named[13386]: socket.c:1111: unexpected error:
Jan 30 22:36:33 stellar named[13386]: internal_send: ::1#33080: Invalid argument
Jan 30 22:36:33 stellar named[13386]: client ::1#33080: error sending response: invalid file
Jan 30 22:36:48 stellar named[13386]: socket.c:1111: unexpected error:
Jan 30 22:36:48 stellar named[13386]: internal_send: ::1#33080: Invalid argument
Jan 30 22:36:48 stellar named[13386]: client ::1#33080: error sending response: invalid file

stellar pri # lsof -i 6 -P -n | grep 33080
named     13379    named   15u  IPv6  58872       UDP *:33080 
named     13384    named   15u  IPv6  58872       UDP *:33080 
named     13386    named   15u  IPv6  58872       UDP *:33080 
named     13387    named   15u  IPv6  58872       UDP *:33080 
named     13388    named   15u  IPv6  58872       UDP *:33080 
Comment 7 Stewart (RETIRED) gentoo-dev 2004-02-01 23:38:42 UTC
This sounds more and more like an upstream bug. Have you been in contact with ISC (even through bind9-users mailing list) about this? If there's a patch available, eg; if it's a known issue, I'll happily apply it to the tree while we wait for a software revision that adresses the issue.
Comment 8 Eric Radman 2004-02-02 10:44:38 UTC
Okay, I'm going to query the BIND mailing list to see if they know anything about this
Comment 9 Eric Radman 2004-02-24 12:50:12 UTC
I'm closing this case because I think it's an upstream bug. After watching the BIND9 mailing list for a while I noticed that there have been many reports of errors in socket.c for various reasons. I suspect this issue will iron itself out with future releases.