|
||||||
| Forums FAQ | Members List | Search | Today's Posts | Mark Forums Read |
| Novell Archives Archived content from Novell openSUSE support forums |
|
|
|
LinkBack | Thread Tools | Display Modes |
|
|||
|
>
> /etc/resolv.conf > -------------------------- > nameserver 212.104.130.9 # ISP nameserver 1 > nameserver 212.104.130.65 # ISP nameserver 2 > nameserver 172.16.1.1 # Vigor 2600We Router > I don't see a "domain" entry in your resolv.conf. domain mysite.com # My domain Not sure if that would create the problem you have. |
|
|||
|
I usually have a domain entry but in my last round of troubleshooting
I had removed it and forgotten to put it back :-) In any case the problems exists with and without the domain entry. I guess the question I am really asking the forum is, How can I force Suse to use TCP instead of UDP for its DNS queries? GofBorg wrote: >>/etc/resolv.conf >>-------------------------- >>nameserver 212.104.130.9 # ISP nameserver 1 >>nameserver 212.104.130.65 # ISP nameserver 2 >>nameserver 172.16.1.1 # Vigor 2600We Router >> > > > I don't see a "domain" entry in your resolv.conf. > > domain mysite.com # My domain > > > Not sure if that would create the problem you have. |
|
|||
|
Daniel de Vries wrote:
> I usually have a domain entry but in my last round of troubleshooting I > had removed it and forgotten to put it back :-) > > In any case the problems exists with and without the domain entry. > > I guess the question I am really asking the forum is, How can I force > Suse to use TCP instead of UDP for its DNS queries? > > > > GofBorg wrote: > >>> /etc/resolv.conf >>> -------------------------- >>> nameserver 212.104.130.9 # ISP nameserver 1 >>> nameserver 212.104.130.65 # ISP nameserver 2 >>> nameserver 172.16.1.1 # Vigor 2600We Router >>> >> >> >> I don't see a "domain" entry in your resolv.conf. >> >> domain mysite.com # My domain >> >> >> Not sure if that would create the problem you have. When a lookup fails, how is the lookup done? Can you do a dig www.ibm.com @212.104.130.9 for example? |
|
|||
|
Issuing the command...
> dig www.ibm.com @212.104.130.9 gives... ------------------ ; <<>> DiG 9.3.1 <<>> www.ibm.com @212.104.130.9 ; (1 server found) ;; global options: printcmd ;; connection timed out; no servers could be reached ------------------ and the following log entry is sent from router to syslog... ------------------ Mar 23 20:07:11 router Vigor: Directed ARP request - ARP request directly directed to a host (not broadcasted) ------------------ The request definitely hits the router, after that I'm not sure. As I said previously, using 'host' instead of 'dig' fails in the same way, except when I use the '-T' option forcing it to use TCP/IP instead of the default UDP. Checking the man pages for dig I couldn't see an equivalent option. Timothy J. Bogart wrote: > Daniel de Vries wrote: > >> I usually have a domain entry but in my last round of troubleshooting >> I had removed it and forgotten to put it back :-) >> >> In any case the problems exists with and without the domain entry.>> >> I guess the question I am really asking the forum is, How can I force >> Suse to use TCP instead of UDP for its DNS queries? >> >> >> >> GofBorg wrote: >> >>>> /etc/resolv.conf >>>> -------------------------- >>>> nameserver 212.104.130.9 # ISP nameserver 1 >>>> nameserver 212.104.130.65 # ISP nameserver 2 >>>> nameserver 172.16.1.1 # Vigor 2600We Router >>>> >>> >>> >>> I don't see a "domain" entry in your resolv.conf. >>> >>> domain mysite.com # My domain >>> >>> >>> Not sure if that would create the problem you have. > > > When a lookup fails, how is the lookup done? Can you do a > > dig www.ibm.com @212.104.130.9 > > for example? |
|
|||
|
Daniel de Vries wrote:
> Timothy J. Bogart wrote: > >> Daniel de Vries wrote: >> >>> I usually have a domain entry but in my last round of troubleshooting >>> I had removed it and forgotten to put it back :-) >>> >>> In any case the problems exists with and without the domain entry. >>> >>> I guess the question I am really asking the forum is, How can I force >>> Suse to use TCP instead of UDP for its DNS queries? >>> >>> >>> >>> GofBorg wrote: >>> >>>>> /etc/resolv.conf >>>>> -------------------------- >>>>> nameserver 212.104.130.9 # ISP nameserver 1 >>>>> nameserver 212.104.130.65 # ISP nameserver 2 >>>>> nameserver 172.16.1.1 # Vigor 2600We Router >>>>> >>>> >>>> >>>> I don't see a "domain" entry in your resolv.conf. >>>> >>>> domain mysite.com # My domain >>>> >>>> >>>> Not sure if that would create the problem you have. >> >> >> >> When a lookup fails, how is the lookup done? Can you do a >> >> dig www.ibm.com @212.104.130.9 >> >> for example? > Issuing the command... > > > dig www.ibm.com @212.104.130.9 > > gives... > > ------------------ > ; <<>> DiG 9.3.1 <<>> www.ibm.com @212.104.130.9 > ; (1 server found) > ;; global options: printcmd > ;; connection timed out; no servers could be reached > ------------------ > > and the following log entry is sent from router to syslog... > > ------------------ > Mar 23 20:07:11 router Vigor: Directed ARP request - ARP request > directly directed to a host (not broadcasted) > ------------------ > > The request definitely hits the router, after that I'm not sure. > > As I said previously, using 'host' instead of 'dig' fails in the same > way, except when I use the '-T' option forcing it to use TCP/IP instead > of the default UDP. Checking the man pages for dig I couldn't see an > equivalent option. > > > > Hmm, I had missed the -T discussion, sorry. In my case, a certain internet cafe just stops responding to DNS requests, but a ping (default ping, UDP) works to the DNS server. Then it just comes back and is fine - then may or may not mysteriously drop out. I don't see this other places. Seemingly there are not complaints from Windows users, but an OSX (BSD) user confirms the behaviour. So, I don't see this as a legitimate security thing, since in both your case and my case it isn't always blocked. And what ever is broken tends to get ignored because 'it works with Windows'. Frustrating. |
|
|||
|
Yeah it is a bit of a pickle. I can't help thinking that there is
something I'm missing as surely more people would be complaining of this problem? I read this morning that as of 2000 a certain proprietary operating system moved to an entirely TCP/IP based DNS implementation which could explain why those clients on my network can perform successful lookups while the Linux hosts fail. Still stumped... Daniel Timothy J. Bogart wrote: > Daniel de Vries wrote: > >> Timothy J. Bogart wrote: >> >>> Daniel de Vries wrote: >>> >>>> I usually have a domain entry but in my last round of >>>> troubleshooting I had removed it and forgotten to put it back :-) >>>> >>>> In any case the problems exists with and without the domain entry. >>>> >>>> I guess the question I am really asking the forum is, How can I >>>> force Suse to use TCP instead of UDP for its DNS queries? >>>> >>>> >>>> >>>> GofBorg wrote: >>>> >>>>>> /etc/resolv.conf >>>>>> -------------------------- >>>>>> nameserver 212.104.130.9 # ISP nameserver 1 >>>>>> nameserver 212.104.130.65 # ISP nameserver 2 >>>>>> nameserver 172.16.1.1 # Vigor 2600We Router >>>>>> >>>>> >>>>> >>>>> I don't see a "domain" entry in your resolv.conf. >>>>> >>>>> domain mysite.com # My domain >>>>> >>>>> >>>>> Not sure if that would create the problem you have. >>> >>> >>> >>> >>> When a lookup fails, how is the lookup done? Can you do a >>> >>> dig www.ibm.com @212.104.130.9 >>> >>> for example? > > > Issuing the command... > > > > > dig www.ibm.com @212.104.130.9 > > > > gives... > > > > ------------------ > > ; <<>> DiG 9.3.1 <<>> www.ibm.com @212.104.130.9 > > ; (1 server found) > > ;; global options: printcmd > > ;; connection timed out; no servers could be reached > > ------------------ > > > > and the following log entry is sent from router to syslog... > > > > ------------------ > > Mar 23 20:07:11 router Vigor: Directed ARP request - ARP request> > directly directed to a host (not broadcasted) > > ------------------ > > > > The request definitely hits the router, after that I'm not sure.> > > > As I said previously, using 'host' instead of 'dig' fails in the same > > way, except when I use the '-T' option forcing it to use TCP/IP instead > > of the default UDP. Checking the man pages for dig I couldn't see an > > equivalent option. > > > > > > > > > > Hmm, I had missed the -T discussion, sorry. > > In my case, a certain internet cafe just stops responding to DNS > requests, but a ping (default ping, UDP) works to the DNS server. Then > it just comes back and is fine - then may or may not mysteriously drop > out. I don't see this other places. Seemingly there are not complaints > from Windows users, but an OSX (BSD) user confirms the behaviour. > > So, I don't see this as a legitimate security thing, since in both your > case and my case it isn't always blocked. And what ever is broken tends > to get ignored because 'it works with Windows'. > > Frustrating. |
|
|||
|
I found the 'use tcp' option for dig and as you could have guessed got
the following result: ------------------------------------------ dig +tcp www.ibm.com @212.104.130.9 ; <<>> DiG 9.3.1 <<>> +tcp www.ibm.com @212.104.130.9 ; (1 server found) ;; global options: printcmd ;; Got answer: ;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 17655 ;; flags: qr rd ra; QUERY: 1, ANSWER: 6, AUTHORITY: 0, ADDITIONAL: 0 ;; QUESTION SECTION: ;www.ibm.com. IN A ;; ANSWER SECTION: www.ibm.com. 143 IN A 129.42.19.99 www.ibm.com. 143 IN A 129.42.20.99 www.ibm.com. 143 IN A 129.42.21.99 www.ibm.com. 143 IN A 129.42.16.99 www.ibm.com. 143 IN A 129.42.17.99 www.ibm.com. 143 IN A 129.42.18.99 ;; Query time: 22 msec ;; SERVER: 212.104.130.9#53(212.104.130.9) ;; WHEN: Thu Mar 24 10:47:32 2005 ;; MSG SIZE rcvd: 125 ------------------------------------------ So really I'm back to my original question... "How can I force Linux to use TCP when querying DNS ?" Daniel Daniel de Vries wrote: > Yeah it is a bit of a pickle. I can't help thinking that there is > something I'm missing as surely more people would be complaining of this > problem? > > I read this morning that as of 2000 a certain proprietary operating > system moved to an entirely TCP/IP based DNS implementation which could > explain why those clients on my network can perform successful lookups > while the Linux hosts fail. > > Still stumped... > > Daniel > > > Timothy J. Bogart wrote: > >> Daniel de Vries wrote: >> >>> Timothy J. Bogart wrote: >>> >>>> Daniel de Vries wrote: >>>> >>>>> I usually have a domain entry but in my last round of >>>>> troubleshooting I had removed it and forgotten to put it back :-) >>>>> >>>>> In any case the problems exists with and without the domain entry. >>>>> >>>>> I guess the question I am really asking the forum is, How can I >>>>> force Suse to use TCP instead of UDP for its DNS queries? >>>>> >>>>> >>>>> >>>>> GofBorg wrote: >>>>> >>>>>>> /etc/resolv.conf >>>>>>> -------------------------- >>>>>>> nameserver 212.104.130.9 # ISP nameserver 1 >>>>>>> nameserver 212.104.130.65 # ISP nameserver 2 >>>>>>> nameserver 172.16.1.1 # Vigor 2600We Router >>>>>>> >>>>>> >>>>>> >>>>>> I don't see a "domain" entry in your resolv.conf. >>>>>> >>>>>> domain mysite.com # My domain >>>>>> >>>>>> >>>>>> Not sure if that would create the problem you have. >>>> >>>> >>>> >>>> >>>> >>>> When a lookup fails, how is the lookup done? Can you do a >>>> >>>> dig www.ibm.com @212.104.130.9 >>>> >>>> for example? >> >> >> > Issuing the command... >> > >> > > dig www.ibm.com @212.104.130.9 >> > >> > gives... >> > >> > ------------------ >> > ; <<>> DiG 9.3.1 <<>> www.ibm.com @212.104.130.9 >> > ; (1 server found) >> > ;; global options: printcmd >> > ;; connection timed out; no servers could be reached >> > ------------------ >> > >> > and the following log entry is sent from router to syslog... >> > >> > ------------------ >> > Mar 23 20:07:11 router Vigor: Directed ARP request - ARP request >> > directly directed to a host (not broadcasted) >> > ------------------ >> > >> > The request definitely hits the router, after that I'm not sure. >> > >> > As I said previously, using 'host' instead of 'dig' fails in the same >> > way, except when I use the '-T' option forcing it to use TCP/IP >> instead >> > of the default UDP. Checking the man pages for dig I couldn't see an >> > equivalent option. >> > >> > >> > >> > >> >> Hmm, I had missed the -T discussion, sorry. >> >> In my case, a certain internet cafe just stops responding to DNS >> requests, but a ping (default ping, UDP) works to the DNS server. >> Then it just comes back and is fine - then may or may not mysteriously >> drop out. I don't see this other places. Seemingly there are not >> complaints from Windows users, but an OSX (BSD) user confirms the >> behaviour. >> >> So, I don't see this as a legitimate security thing, since in both >> your case and my case it isn't always blocked. And what ever is >> broken tends to get ignored because 'it works with Windows'. >> >> Frustrating. |
|
|||
|
> So really I'm back to my original question...
> > "How can I force Linux to use TCP when querying DNS ?" You wouldn't have a Firewall involved here would you? Perhaps the SuSE firewall? I know some firewalls are configured to dump UDP queries. I ran into this problem myself when working with DNS from my Linux server. My solution was to open a port for the reverse UDP transaction. Then all was well. I don't know of a way to force one protocol over the other. |
|
|||
|
GofBorg wrote:
>>So really I'm back to my original question... >> >>"How can I force Linux to use TCP when querying DNS ?" > > > You wouldn't have a Firewall involved here would you? > Perhaps the SuSE firewall? > > I know some firewalls are configured to dump UDP queries. > I ran into this problem myself when working with DNS from > my Linux server. My solution was to open a port for the reverse > UDP transaction. Then all was well. I don't know of a way to force> one protocol over the other. > > Yeah I that occurred to me too. I don't have the Suse firewall enabled by default and even with the routers built in firewall and 'DOS' protection disabled the problem still exists. :-( |
|
|||
|
Daniel de Vries wrote:
I haven't seen question about 3 entries in your /etc/resolv.conf. Your router address is last entry. Try to put it first, or to make it the only choice. Have you tried: dig www.ibm.com @172.16.1.1 -- Greetings, Rajko |
|
| Bookmarks |
| Thread Tools | |
| Display Modes | |
|
|