nslookup hangs

[Bob Schmid]

Can anyone tellme why nslookup is used in the engine when starting up threads…irregardless, it seems, as to netsvc.conf file specified as far as search order as local,bind.

The engine began hanging Saturday night due to -> nslookoup hangs <- it uses the "homegrown host names from NetConfig, to search DNS even though we are saying to look at /etc/hosts first. Obbviously nslookup does not utilize the netsvc.conf specified search chain…and obvisouly our hostnames define in /etc/hosts are NOT defined in teh DNS servers…thus the hang I presume….but why is this showing itself suddenly on our AIX boxes WHERE NO CHANGES HAVE OCCURRED! HELP!

Topic

[Charlie Bursell]

I added that Bob.   It was added because in the past NetConfig would only look at the hosts file if you provided an non-dotted address.  If you were using DNS you would get a warning message “Host Not Found”.  I added the code that if the address was not found in the host file to look it up using nslookup.

You may have an older version since we did have some problems with it on some machines.  For example, WIN 2003 gave us problems prrior to Tcl 8.4 with any exec call.  You are certainly free to comment it out if it gives you problems.  Look in the file $HCIROOT/tcl/lib/cloverleaf/netData.tlib

If you have an older version of CL maybe a newer version would work better 🙂

[Bob Schmid]

just showed up as  aproblem Saturday night ….? 😯

[Bob Schmid]

this definitely has never been defined on the dns servers.

…only in /etc/hosts…..than “suddently” Saturday night…..boom crash bam….nslookup is hanging looking for it out on the DNS server ?????

if you help me here…..I’ll personally fly out there and give you a big kiss….and believe me….that is really stretching my wow factor equation ! 😳

[Charlie Bursell]

Can you run nslookup from the command line?  If it hangs there it will likely hang in CL.  As I said, for a quick work-around just comment it out.

[Bob Schmid]

right….it hangs……could my hosts file be messed up ?….when I execute “host” or traceroute…they do work.

[Bob Schmid]

Charlie on Satyurday …I changed all my NetCOnfig entries with hardcoded IP’s

[John Mercogliano]

Robert,

  If your nslookup failed then something might be happening with your dns server you are communicating with.  This would cause problems with any application that needs to lookup an address not just cloverleaf.  You might want to check the last modified dates on the /etc/*.conf files especially resolv.conf to see if something changed on your system.  

  Are you the administrator or do you have a separate group?  If you still can’t run nslookup and nothing has changed on your system, then you need to contact your group that is in charge of your DNS servers.  

  Now you also, said you ran traceroute, so check to see if any patches have been loaded.  At least on Linux, I’ve had some experience with a missing file in a patch for malloc that caused nslookup to fail but other commands that resolved names worked fine.  

  Hopefully, this is not to much of a needle in a haystack problem for you.

Good Luck,

John Mercogliano
Sentara Healthcare
Hampton Roads, VA

[Bob Schmid]

Here is what I have found out since this morning:

Per Charlie Bursell on Clovertech, code in netData.tndx was implemented in 5.6 that had the engine doing nslookups at engine startup before looking at

[Russ Ross]

Sorry to hear about your DNS server issues.

I had to give up on our DNS servers long ago because they can’t keep the DNS servers working much of the time.

Things aren’t getting better becuase now not every machine is even being defined in our DNS server which is a head scratcher to me but security is starting to take a stance you should now the machine exists so my nslookups no longer even find servers on our in-house network.

Sounds like I might be getting some hidden benifit as well when cloverleaf might want to force me to use our unreliable DNS servers.

Ideally hard coding of IPs seems like the poorer choice but our reality has shown it to be the more robust and maintanable choice.

So don’t dispair, we are proof that life goes on with hard coded IPs.

However, we do use the /etc/hosts file sometimes and depend on it being looked at first so now that is a bit of a question mark, too.

Having said that, what you described brought to mind a related problem I had with getting my xterms to launch because I use was using the hostname of my cloverleaf servers to launch the xterm logins, which weren’t getting resolved.

As Charlie already pointed out the /etc/*.conf files can play a role and Charlie’s comment is what triggered my memory for this reply.

In my case the network group had added 2 new domains and not told anyone.

I finally resolved the problem by adding a search entery into my /etc/resolv.conf file so the 2 new domains could be resolved.

Here is what is currently in my /etc/resolv.conf file:

nameserver  143.111.64.254

nameserver  143.111.89.254

nameserver  143.111.201.254

search mdacc.tmc.edu utmdacc.org mdanderson.edu

Russ Ross
RussRoss318@gmail.com

[Bob Schmid]

removing the nslookup code used in netData.tlib got us around the issue.

WHy the dns server was so slow in responding…I may never find out.

Thanks out to everybody for your precious time and tremendous help. (and encouragement ) ❗ ) This was a rough one…and really exploited the

weakness we have in downtime notification throughout our enterprise!

Bob

[Charlie Bursell]

Russ

Not to worry. The /etc/hosts file is always checked first

You can look at the script and verify if you like

[Author]

Replies