[Okta Webinar] Learn how to a build a cloud-first strategyRegister Now

x
  • Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 1210
  • Last Modified:

Solaris 8 Delayed Login Using XDMCP

Hi All

I am having problems trouble shooting a Solaris 8 login problem. I am not a Solaris expert but can hold my own, so bear with me while I do my best to explain.

Two Solaris servers, one is much older and less powerful than the other. The more powerful of the two is having very slow login times from a remote application called Exceed (Hummingbird).

Now first off I should point out I have been down the route of contacting Exceed support and they believe this not be a problem with their software and I am inclined to believe them.

The server with the very slow login using Exceed is Server1. The other machine is called Server2. Server2 is ok with login using Exceed, even though it is a much older, slower machine. I can login to the remote host Server2 on the Server1 using the 'remote login' procedure, and the performance is ok, but again when logging in using the Exceed application the login is very slow.

Server1 via Server2: time to login window - 45 seconds, from login window to full CDE - hangs, > 5 minutes
Server2 via Server1: time to login window - 25 seconds, from login window to full CDE - 90 seconds

Trouble shooting:

I can ping both servers by name and IP address
I have used XDMCP logging and found no errors (logs enclosed)
I have checked, disk space, CPU, network utilisation etc.
I have added the server to the hosts file
The server is located physically on the same LAN (same server room)
The delay is only on logging in, not when logged in
Version of Solaris - SunOS Server1 5.8 Generic_108528-15 sun4u sparc SUNW,Ultra-4
Server1%2

If anyone has any ideas would be very grateful as rather stuck.
0
explorersup
Asked:
explorersup
  • 5
  • 3
1 Solution
 
Brian UtterbackPrinciple Software EngineerCommented:
Delays like this typically come either from name service lookups or NFS mounts. Whenever you login over the network, the system does a reverse lookup on the IP address if the client you are using. It is often the case that this lookup can take a long time if the name service has not been set up right.

After that, the system has to lookup your account info. And depending on where the user's home directory is located, it might require using an NFS mount to get it.

Does it still take a long time if you login once, immediately exit and then login a second time right away? Does the second login happen quickly or take a long time as well?
0
 
explorersupAuthor Commented:
Hi

Thanks for your input.

I'm doing the login, logout, log back in test now. Takes a little while so will report back.

If it is NFS mounts, how can I test/prove this?

I originally thought it may be the lookups, hence the ping's on names and IP address and adding server names into host file.

Is there another test you can suggest to make sure lookup is working ok?

Thanks
0
 
explorersupAuthor Commented:
Hi

Login, logout, log back in test took just as long unfortunately.

Any suggestions?

0
What does it mean to be "Always On"?

Is your cloud always on? With an Always On cloud you won't have to worry about downtime for maintenance or software application code updates, ensuring that your bottom line isn't affected.

 
Brian UtterbackPrinciple Software EngineerCommented:
There are a couple of ways to see where the problem is. Perhaps the easiest is to login to the server as root and then observe what is happening with the ps and ptree commands. At some point a process will stop and wait for something. If you can catch it while it is stopped, you can use the pstack command to see what it is doing.

The method that is guaranteed to find the problem, but is the most work if to use the truss command. You can capture all of the system calls of all the processes involved, but analyzing the data can be tedious and difficult. Better to try the above method first.
0
 
explorersupAuthor Commented:
Thanks for the suggestion.

Im going to try that and see how I get on.

I will report back findings before accepting solution.

Thanks.
0
 
shellghostCommented:
Just a thought, have you checked or compared the /etc/nsswitch.conf file?  If the faster server is set to use local files (i.e. the hosts file) first and the slower one DNS first, that could explain it.  I know you said that you put the server info in the hosts file, I would double check that (both servers have the other servers info in their hosts file) and then check the nsswitch.conf and make sure they are both set to lookup via local files first, then go out to DNS if not found locally.  Good luck!
0
 
explorersupAuthor Commented:
Hi Guys

Thanks for the suggestions, I am still working on this problem and will report back shortly.

0
 
explorersupAuthor Commented:
Hi Guys,

the /etc/nsswitch.conf file is set up as described and the hosts file is also correct. Still having no luck with this one
0
 
Brian UtterbackPrinciple Software EngineerCommented:
Did you try my suggestions about using ptree and truss?
0

Featured Post

Industry Leaders: We Want Your Opinion!

We value your feedback.

Take our survey and automatically be enter to win anyone of the following:
Yeti Cooler, Amazon eGift Card, and Movie eGift Card!

  • 5
  • 3
Tackle projects and never again get stuck behind a technical roadblock.
Join Now