xen discuss - Aug 2006 - DNS resolutions is not working in a Solaris 10 DomU

Hi,

I found a problem with the DNS resolution in a Solaris 10 DomU.

The Test Environment was:

I installed the Xen bfu on top of Solaris snv43 and booted the Dom0 with 
Solaris. Then I installed and booted a Solaris 10 DomU.


In the Solaris 10 Dom0 the DNS resolution works; in the Solaris 10 DomU 
the DNS resolution does not work as expected:


Inside the Solaris 10 DomU I can''t use hostnames resolved by the DNS,
e.g.

bash-3.00# ping www.heise.de
ping: unknown host www.heise.de

The name resolution via /etc/hosts works:

bash-3.00# tail -1 /etc/hosts
152.2.210.80 ibiblio.org

bash-3.00# ping ibiblio.org
ibiblio.org is alive


nslookup works also:

bash-3.00# nslookup www.heise.de
Server:         192.168.178.1
Address:        192.168.178.1#53

Non-authoritative answer:
Name:   www.heise.de
Address: 193.99.144.85


And I can ping the machine via it''s IP address:

bash-3.00# ping 193.99.144.85
193.99.144.85 is alive


The /etc/nsswitch.conf and /etc/resolv.conf are okay:

bash-3.00# egrep "hosts|ipnodes" /etc/nsswitch.conf | grep -v
"^#"
hosts:      files dns   
ipnodes:   files dns


bash-3.00# cat /etc/resolv.conf
domain isbs.de
nameserver 192.168.178.1


Because I can reach the machines via there IP address the IP and routing 
configuration is also okay.

A truss on ping gives the following results:


bash-3.00# truss ping www.heise.de
execve("/usr/sbin/ping", 0x08047E84, 0x08047E90)  argc = 2
resolvepath("/usr/lib/ld.so.1", "/lib/ld.so.1", 1023) = 12
resolvepath("/usr/sbin/ping", "/usr/sbin/ping", 1023) = 14
sysconfig(_CONFIG_PAGESIZE)                     = 4096
xstat(2, "/usr/sbin/ping", 0x08047C48)          = 0
open("/var/ld/ld.config", O_RDONLY)             Err#2 ENOENT
xstat(2, "/lib/libxnet.so.1", 0x08047448)       = 0
resolvepath("/lib/libxnet.so.1", "/lib/libxnet.so.1", 1023)
= 17
open("/lib/libxnet.so.1", O_RDONLY)             = 3
mmap(0x00010000, 4096, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_ALIGN, 3, 0) 
= 0xC6FC0000
mmap(0x00001000, 8192, PROT_NONE, 
MAP_PRIVATE|MAP_NORESERVE|MAP_ANON|MAP_ALIGN, -1, 0) = 0xC6FB0000
mmap(0xC6FB0000, 5292, PROT_READ|PROT_WRITE|PROT_EXEC, 
MAP_PRIVATE|MAP_FIXED|MAP_TEXT, 3, 0) = 0xC6FB0000
mmap(0x00000000, 4096, PROT_READ|PROT_WRITE|PROT_EXEC, 
MAP_PRIVATE|MAP_ANON, -1, 0) = 0xC6FA0000
close(3)                                        = 0
xstat(2, "/lib/libsocket.so.1", 0x08047448)     = 0
resolvepath("/lib/libsocket.so.1", "/lib/libsocket.so.1",
1023) = 19
open("/lib/libsocket.so.1", O_RDONLY)           = 3
mmap(0xC6FC0000, 4096, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_FIXED, 3, 0) 
= 0xC6FC0000
mmap(0x00010000, 114688, PROT_NONE, 
MAP_PRIVATE|MAP_NORESERVE|MAP_ANON|MAP_ALIGN, -1, 0) = 0xC6F80000
mmap(0xC6F80000, 44258, PROT_READ|PROT_EXEC, 
MAP_PRIVATE|MAP_FIXED|MAP_TEXT, 3, 0) = 0xC6F80000
mmap(0xC6F9B000, 2613, PROT_READ|PROT_WRITE, 
MAP_PRIVATE|MAP_FIXED|MAP_INITDATA, 3, 45056) = 0xC6F9B000
munmap(0xC6F8B000, 65536)                       = 0
memcntl(0xC6F80000, 11716, MC_ADVISE, MADV_WILLNEED, 0, 0) = 0
close(3)                                        = 0
xstat(2, "/lib/libnsl.so.1", 0x08047448)        = 0
resolvepath("/lib/libnsl.so.1", "/lib/libnsl.so.1", 1023) =
16
open("/lib/libnsl.so.1", O_RDONLY)              = 3
mmap(0xC6FC0000, 4096, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_FIXED, 3, 0) 
= 0xC6FC0000
mmap(0x00010000, 643072, PROT_NONE, 
MAP_PRIVATE|MAP_NORESERVE|MAP_ANON|MAP_ALIGN, -1, 0) = 0xC6EE0000
mmap(0xC6EE0000, 529981, PROT_READ|PROT_EXEC, 
MAP_PRIVATE|MAP_FIXED|MAP_TEXT, 3, 0) = 0xC6EE0000
mmap(0xC6F72000, 20397, PROT_READ|PROT_WRITE, 
MAP_PRIVATE|MAP_FIXED|MAP_INITDATA, 3, 532480) = 0xC6F72000
mmap(0xC6F77000, 22816, PROT_READ|PROT_WRITE, 
MAP_PRIVATE|MAP_FIXED|MAP_ANON, -1, 0) = 0xC6F77000
munmap(0xC6F62000, 65536)                       = 0
memcntl(0xC6EE0000, 56996, MC_ADVISE, MADV_WILLNEED, 0, 0) = 0
close(3)                                        = 0
xstat(2, "/lib/libm.so.2", 0x08047448)          = 0
resolvepath("/lib/libm.so.2", "/lib/libm.so.2", 1023) = 14
open("/lib/libm.so.2", O_RDONLY)                = 3
mmap(0xC6FC0000, 4096, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_FIXED, 3, 0) 
= 0xC6FC0000
mmap(0x00010000, 364544, PROT_NONE, 
MAP_PRIVATE|MAP_NORESERVE|MAP_ANON|MAP_ALIGN, -1, 0) = 0xC6E80000
mmap(0xC6E80000, 284148, PROT_READ|PROT_EXEC, 
MAP_PRIVATE|MAP_FIXED|MAP_TEXT, 3, 0) = 0xC6E80000
mmap(0xC6ED5000, 14760, PROT_READ|PROT_WRITE|PROT_EXEC, 
MAP_PRIVATE|MAP_FIXED|MAP_INITDATA, 3, 282624) = 0xC6ED5000
munmap(0xC6EC6000, 61440)                       = 0
memcntl(0xC6E80000, 22716, MC_ADVISE, MADV_WILLNEED, 0, 0) = 0
close(3)                                        = 0
xstat(2, "/lib/libinetutil.so.1", 0x08047448)   = 0
resolvepath("/lib/libinetutil.so.1",
"/lib/libinetutil.so.1", 1023) = 21
open("/lib/libinetutil.so.1", O_RDONLY)         = 3
mmap(0xC6FC0000, 4096, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_FIXED, 3, 0) 
= 0xC6FC0000
mmap(0x00000000, 4096, PROT_READ|PROT_WRITE|PROT_EXEC, 
MAP_PRIVATE|MAP_ANON, -1, 0) = 0xC6E70000
mmap(0x00010000, 81920, PROT_NONE, 
MAP_PRIVATE|MAP_NORESERVE|MAP_ANON|MAP_ALIGN, -1, 0) = 0xC6E50000
mmap(0xC6E50000, 8963, PROT_READ|PROT_EXEC, 
MAP_PRIVATE|MAP_FIXED|MAP_TEXT, 3, 0) = 0xC6E50000
mmap(0xC6E63000, 372, PROT_READ|PROT_WRITE, 
MAP_PRIVATE|MAP_FIXED|MAP_INITDATA, 3, 12288) = 0xC6E63000
munmap(0xC6E53000, 65536)                       = 0
memcntl(0xC6E50000, 2760, MC_ADVISE, MADV_WILLNEED, 0, 0) = 0
close(3)                                        = 0
xstat(2, "/lib/libc.so.1", 0x08047448)          = 0
resolvepath("/lib/libc.so.1", "/lib/libc.so.1", 1023) = 14
open("/lib/libc.so.1", O_RDONLY)                = 3
mmap(0xC6FC0000, 4096, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_FIXED, 3, 0) 
= 0xC6FC0000
mmap(0x00010000, 897024, PROT_NONE, 
MAP_PRIVATE|MAP_NORESERVE|MAP_ANON|MAP_ALIGN, -1, 0) = 0xC6D70000
mmap(0xC6D70000, 794871, PROT_READ|PROT_EXEC, 
MAP_PRIVATE|MAP_FIXED|MAP_TEXT, 3, 0) = 0xC6D70000
mmap(0xC6E43000, 23778, PROT_READ|PROT_WRITE, 
MAP_PRIVATE|MAP_FIXED|MAP_INITDATA, 3, 798720) = 0xC6E43000
mmap(0xC6E49000, 5152, PROT_READ|PROT_WRITE, 
MAP_PRIVATE|MAP_FIXED|MAP_ANON, -1, 0) = 0xC6E49000
munmap(0xC6E33000, 65536)                       = 0
memcntl(0xC6D70000, 120792, MC_ADVISE, MADV_WILLNEED, 0, 0) = 0
close(3)                                        = 0
mmap(0x00000000, 4096, PROT_READ|PROT_WRITE|PROT_EXEC, 
MAP_PRIVATE|MAP_ANON, -1, 0) = 0xC6D60000
munmap(0xC6FC0000, 4096)                        = 0
mmap(0x00010000, 24576, PROT_READ|PROT_WRITE|PROT_EXEC, 
MAP_PRIVATE|MAP_ANON|MAP_ALIGN, -1, 0) = 0xC6D50000
getcontext(0x080479F0)
getrlimit(RLIMIT_STACK, 0x080479E8)             = 0
getpid()                                        = 103279 [103278]
lwp_private(0, 1, 0xC6D52000)                   = 0x000001C3
setustack(0xC6D52060)
sigfillset(0xC6E49578)                          = 0
sysi86(SI86FPSTART, 0xC6E49AEC, 0x0000133F, 0x00001F80) = 0x00000001
getuid()                                        = 0 [0]
getuid()                                        = 0 [0]
fstat64(1, 0x08047CF0)                          = 0
getpid()                                        = 103279 [103278]
sysconfig(_CONFIG_PAGESIZE)                     = 4096
open("/etc/netconfig", O_RDONLY)                = 3
fstat64(3, 0x08047580)                          = 0
brk(0x0809BE78)                                 = 0
brk(0x0809FE78)                                 = 0
fstat64(3, 0x080474A0)                          = 0
ioctl(3, TCGETA, 0x0804754C)                    Err#25 ENOTTY
read(3, " #   C D D L   H E A D E".., 8192)     = 2150
read(3, 0x0809BE84, 8192)                       = 0
llseek(3, 0, SEEK_CUR)                          = 2150
llseek(3, 0, SEEK_SET)                          = 0
read(3, " #   C D D L   H E A D E".., 8192)     = 2150
read(3, 0x0809BE84, 8192)                       = 0
llseek(3, 0, SEEK_CUR)                          = 2150
close(3)                                        = 0
open("/dev/udp", O_RDONLY)                      = 3
ioctl(3, SIOCGLIFNUM, 0x08047A64)               = 0
close(3)                                        = 0
open("/dev/udp6", O_RDONLY)                     = 3
ioctl(3, SIOCGLIFNUM, 0x08047A84)               = 0
close(3)                                        = 0
open("/dev/udp", O_RDONLY)                      = 3
ioctl(3, SIOCGLIFNUM, 0x08047A84)               = 0
close(3)                                        = 0
brk(0x0809FE78)                                 = 0
brk(0x080A1E78)                                 = 0
open64("/var/run/name_service_door", O_RDONLY)  = 3
fcntl(3, F_SETFD, 0x00000001)                   = 0
door_info(3, 0xC6E491F0)                        = 0
door_call(3, 0x080459F8)                        = 0
door_info(3, 0x080459C0)                        = 0
door_call(3, 0x080459F8)                        = 0
fstat64(2, 0x08046E20)                          = 0
pingwrite(2, " p i n g", 4)                             = 4
: unknownwrite(2, " :   u n k n o w n", 9)              = 9
 host write(2, "   h o s t  ", 6)                       = 6
www.heise.dewrite(2, " w w w . h e i s e . d e", 12)    = 12

write(2, "\n", 1)                               = 1
_exit(1)
bash-3.00#


While doing the "ping www.heise.de" I can not see any traffic between 
the DomU
and the nameserver via snoop in the Dom0.

It this a known bug?

regards

Bernd



-- 
Bernd Schemmer, Frankfurt am Main, Germany
http://home.arcor.de/bnsmb/index.html

Más temprano que tarde el mundio cambiará
                        Fidel Castro

On Tue, Aug 01, 2006 at 11:49:07PM +0000, Bernd Schemmer wrote:
> bash-3.00# ping www.heise.de
> ping: unknown host www.heise.de
> nslookup works also:
This is odd, but it''s not obvious how it can be Xen related. What does

getent hosts www.heise.de say?

Do you get anything interesting in nscd.log if you enable it (/etc/nscd.conf)?

Does restarting nscd help at all?

regards
john

Bernd Schemmer wrote:

fyi:  The problem also exists in a Solaris DomU created with minimal 
necessary number of packages. The problem does not exist in a Linux DomU.

regards

Bernd

-- 
Bernd Schemmer, Frankfurt am Main, Germany
http://home.arcor.de/bnsmb/index.html

Más temprano que tarde el mundio cambiará
                        Fidel Castro

>Very strange - now it works. I did not change anything since yesterday
>-only shutdown the Laptop and rebooted it today.
I''ve seen this sort of thing occur when there''s been TCP
checksum
offloading problems...

John Levon wrote:

John,
>getent hosts www.heise.de say?
>  
>
bash-3.00# getent hosts www.heise.de  
193.99.144.85   www.heise.de    
     
bash-3.00# ping www.heise.de          
www.heise.de is alive                 

Very strange - now it works. I did not change anything since yesterday 
-only shutdown the Laptop and rebooted it today.

Thanks

Bernd



-- 
Bernd Schemmer, Frankfurt am Main, Germany
http://home.arcor.de/bnsmb/index.html

Más temprano que tarde el mundio cambiará
                        Fidel Castro

I''m seeing the same thing in a new install of Solaris build-81 with
Solaris dom0 and domU.  Pointing a Solaris domU to a known good DNS server
(works from dom0) fails to resolve anything.  Network connectivity seems OK; I
can ping the DNS server, but I get timeouts resolving hostnames.
 
 
This message posted from opensolaris.org

Chris wrote:
> I''m seeing the same thing in a new install of Solaris build-81
with Solaris dom0 and domU.  Pointing a Solaris domU to a known good DNS server
(works from dom0) fails to resolve anything.  Network connectivity seems OK; I
can ping the DNS server, but I get timeouts resolving hostnames.
>  
>  
> This message posted from opensolaris.org
> _______________________________________________
> xen-discuss mailing list
> xen-discuss@opensolaris.org
>
>   
Did you set the /etc/system entry to disable the  TCP checksum 
(shouldn''t be necessary anymore but who knows)?

I did not install snv81 yet but will install it in one of the next days 
(in Dom0 and DomU)

regards

Bernd


-- 
Bernd Schemmer, Frankfurt am Main, Germany
http://home.arcor.de/bnsmb/index.html

M s temprano que tarde el mundo cambiar .
                        Fidel Castro

TCP checksum was indeed causing the problem.  Adding this line to /etc/system in
the domU solves the problem:

set xnf:xnf_cksum_offload = 0

Thanks!
 
 
This message posted from opensolaris.org

Hi,
>>TCP checksum was indeed causing the problem.  Adding this line to
/etc/system in the domU solves the problem:
>>set xnf:xnf_cksum_offload = 0

Good, but strange

in a Solaris snv_81 DomU running in a Solaris  snv_78 Dom0 the name 
resolution works without the /etc/system entry.

bash-3.2# getent hosts www.heise.de
193.99.144.85   www.heise.de


regards

Bernd


Chris wrote:
> TCP checksum was indeed causing the problem.  Adding this line to
/etc/system in the domU solves the problem:
>
> set xnf:xnf_cksum_offload = 0
>
> Thanks!
>  
>  
> This message posted from opensolaris.org
> _______________________________________________
> xen-discuss mailing list
> xen-discuss@opensolaris.org
>
>   

-- 
Bernd Schemmer, Frankfurt am Main, Germany
http://home.arcor.de/bnsmb/index.html

M s temprano que tarde el mundo cambiar .
                        Fidel Castro

> I''m seeing the same thing in a new install of Solaris
> build-81 with Solaris dom0 and domU.  Pointing a
> Solaris domU to a known good DNS server (works from
> dom0) fails to resolve anything. 
The DNS server is external to dom0, that is, we have to
go over a wire to reach the DNS server?  Or is the DNS
server running on the dom0?

What kind of NIC is used on the opensolaris build81 dom0?
> Network
> connectivity seems OK; I can ping the DNS server, but
> I get timeouts resolving hostnames.
...
> TCP checksum was indeed causing the problem.
> Adding this line to /etc/system in the domU solves the problem:
>
> set xnf:xnf_cksum_offload = 0

Would it be possible that you run a "tcpdump -v -s 0"
on the DNS server or some router between dom0 and the
DNS server, to verify that there really is an issue with
the dom0/domU sending out packes with bad UDP or
TCP checksums?
 
 
This message posted from opensolaris.org