I'm having some problems with our HP-UX environment. I'm running Oracle EBS on a HP-UX system, and whenever my primary dns server goes down some of my clients can no longer connect. Everything on my windows side continues to work, and clients can reach the internet and other services without a problem, so my best guess is that it is only a problem with my HP-UX system.
I have a secondary dns server defined so I'm not sure why it's not failing over? Is there a timeout issue, or something else I'm missig here? Please see below for how my conf files look like.
Difficult to help with the little info you give, we have no real idea of your network topology nor how your win pc are connected and configured...And whats in your DNS servers...
Is every thing on same net or do you have subnets?...
When you say clients can reach internet and other services, what are the other services? on the same network?....
My bet is that the defaults mean that the client connection as a whole times out before the DNS failure is noticed and redirected to the secondary DNS. An incoming request is just an IP address, so your server will try to resolve that to a name for you.
Have a closer look at the manual for resolv.conf
The defaults will mean a 20 second timeout, however this might be extended if you have to look up the name of the DNS server than does respond, and that goes to the first one again.
For local sessions, you can also set the environment variables RES_RETRANS & RES_RETRY to something more like 1000 & 1 respectively.
I hope that this helps. It might not be the cause, but it's something we have hit in the past.
I.e. your resolve.conf settings (provided it works okay on your HP-UX) must also exist on each PC, either static or provided by the DHCP server.
And your DNS must be okay:
nslookup -q=ns domain.local
must give both name servers.
The PCs might use WINS as a fallback if DNS is not working.
Assuming that 192.168.1.1 and 192.168.1.2 are both DNS servers and that they are Primary and Secondary respectively.
You can test whether they are both working DNS servers with:
nslookup microsoft.com 192.168.1.1
nslookup microsoft.com 192.168.2.1
The /etc/nsswitch.conf posted will give up if the Primary DNS is down.
With "NOTFOUND=return" first I don't think that your server will ever look at the secondary (or any reserve) DNS's listed in your /etc/resolv.conf .
Unless your server does tons of DNS lookups on computers which are for some reason not in your DNS, imho this is much better:
hosts: dns [NOTFOUND=continue UNAVAIL=continue TRYAGAIN=continue] files
(I cut/paste this from one of my long-term reliable computers. Not my original work and I would like to credit the tip, but it's been too long).
No, it's safer to have files first!
Especially "localhost" should not be searched in an external name service.
E.g. "swinstall" and several applications talk to localhost. And should work even if the network cable is unplugged...