Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations gkittelson on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

Problem with name resolution 1

Status
Not open for further replies.

zaxxon

MIS
Dec 12, 2001
226
DE
Hello,

we updated the resolv.conf and /etc/hosts on several AIX boxes from AIX 4.3.3 to 5.3.
When I do a nslookup on a hostname or ip-adress, it automatically takes one of the new nameservers given in the updated resolv.conf. Our netsvc.conf order is "local,bind".
The strange thing is, that the old DNS server still gets the requests from our updated boxes.
We already tried to user namerslv -e to feed the system configuration database (ODM?..), named in the man page for namerslv.
We then had a tcpdump running and grep'ing and still saw requests to the old DNS server.
Does anybody know how to refresh or reload that kind of "caching" behaviour?
We want to prevent to restart all applications or worse to reboot all boxes just to have the resolving being corrected.

At the moment both the old and the new DNS servers are available but there will be some point in the future, where the old DNS servers will be turned off.
We think that when the boxes don't find the old DNS servers anymore, they will use the new ones by doing gethostbyname() for example and so automatically following the order in /etc/netsvc.conf, finding the new config...

Any hints are welcome.

laters
zaxxon
 
@DukeSSD
There is no /etc/irs.conf on any of the machines. I will check the link you posted, thanks.

@Khalidaaa
No because I don't think it has anything to do with it. It uses maybe gethostbyname() like any other application but it has nothing to do which way the name resolving is handled.
But nevertheless I restarted it instead of refreshing to be sure on one box and it still shows up in the DNS server's logs. Thanks though.

laters
zaxxon
 
Still no solution yet so *bump* on the thread, sorry.

laters
zaxxon
 
You could try and see what happens on the "nameserver clients" when you unplug the old DNS server from the network (on a low-impact time late at night perhaps?).


HTH,

p5wizard
 
What is the content of your file /etc/resolv.conf? BTW, our AIX 5.3 machine has local,bind4 (with a 4) in the netsvc.conf file.

If you have multiple nameserver entries in the resolv.conf file, the OS will try to contact the primary server up to three times, I believe. If that failed, the OS will try the next server in the list. So, if your old nameserver still is in the resolv.conf, there is a possibility that it may be used.

At the command line try typing "nslookup" and it will display what primary name server your machine is using.

In the interactive "nslookup" mode, you can tell the OS what name server you'd like to use and can test to see if that name server actually responds to query. Something like:
nslookup
server <IP of server>
<host name to look up>
exit
 
BTW, how did you determine that "the old DNS server still gets the requests from our update boxes"? Are your name servers Unix boxes or windows, or what?
 
- We can use the new nameservers without any problems, using host or nslookup and so on. But something is still talking the old nameservers on a bunch of boxes with many different applications on it.
- Our DNS-servers are linux boxes. We can see the requests of the "clients" in the log it writes, also with tcpdump.
- We can't do a refresh on the named on the clients because those boxes have no named running.
- Our /etc/resolv.conf looks like the documentation wants it, where the nameserver IPs are of course the new DNS-servers.
Code:
nameserver 10.10.10.10
nameserver 10.10.100.100
search domain1.de domain2.de domain3.de domain4.de

We tried it with terminating and starting sendmail, no luck.
It seems someone has to decide to reboot all boxes (doh!) or just sit and wait and look what will happen if the old DNS-servers go offline.

laters
zaxxon
 
I'd try with

hosts = local4,bind4

in netsvc.conf on the AIX53 clients, to force use TCP v4 -

I'd just like to see/know if it makes a difference.


HTH,

p5wizard
 
Hi

Could it be due to ro records TTL (Time To Live)
If i.e. you have 30 Days in TTL, then the local systems cache the records in 30 days before resolving again.

/hga
 
Will check with local4,bind4 asap. TTL for tcpip or.. ? I will check it too.


laters
zaxxon
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top