Problem with name resolution 1

[slgordon]

I am trying to setup an RS/6000 to look to my NT DNS servers to resolve names. However, as soon as I add the nameserver and the domain name to the resolv.conf file,
it take approx 30-60 secs to telnet to it (ie When you go to a DOS prompt and type in telnet "ip" it just sits there for 30-60 secs) Once I get into the box, it runs fine. I have also updated our DNS/WINS with the IP of the new RS/6000.
Please help before I throw myself off a building!
Thanks in advance!

 

[PSD]

slgordon,

Can I assume that you are using an /etc/netsvc.conf file?

Normally I specify that the machine uses local i.e. /etc/hosts first. If you select bind4 first and the NT DNS server is not pingable on reboot of the RS then it will hang.

hosts=local,bind4

I use bind4 because in AIX 4.3.0 and beyond the bind daemon looks for IP version 6 addresses which can slow down name resolution through nslookup and also via telnet. This could be the problem - try the following in the /etc/netsvc.conf file:

hosts=bind4,local

If this sorts the problem out then it is a IPV6 issue, i.e. your NT DNS servers are saying to the RS/6000 request "I do not have a IPV6 address for this host" after a while it times out and says "here is a IPV4 address"....This is a known problem and is documented in APAR IX76289.

Let me know how you get on....

PSD
HACMP Specialist

 

[slgordon]

Yeah, I was using the the netsvc.conf file, but I was using hosts=local,bind. Once I changed it to "bind4" it works great!
Thanks for you help!

 

[slgordon]

I also want to change one of our boxes that is using HACMP to use name resolution, will that make a difference?

 

[PSD]

slogordon,

You can set then up to use DNS but you need to also change the run time parameters within HACMP i.e.

smitty cm_cfg_res_menu

change / show run time param.

select each node one at a time and change the field:

host use NIS or name server = true

The reason for this is:

"If the cluster uses NIS or a Name Server, this field shouldbe set to "true". This will allow the scripts to turn on/off these services when they would interfere with reconfiguration i.e. failover.

I would then the next time you stop HACMP sync. cluster resources from the cluster node you have changed to propogate this out to the other nodes.

Hope that this helps

Cheers

PSD
HAMCP Specialist

 

[cspilman]

You won't necessarily have to stop HACMP to resync the cluster resources. If you have DARE (Dynamic Automatic Reconfiguration Events) running, you can resync the cluster without stopping the cluster. Unfortunately I do not remember what versions this applies to.

Good Luck!
Chuck

 

[slgordon]

OK...thanks for all of your help!

 

[PSD]

Sglordon,

You can sync. cluster resources on the fly i.e. through DARE but I personally do most things offline where possible - I have had some strange things happen when using DARE - if you are going to use it make sure that you are on HAMCP 4.3.1 patched up or HAMCP 4.4 patched up.

Good luck

PSD