(Summary) Re: problem with rlogin on solaris 2.4

From: Peter Kim (peterk@jdc.csg.mot.com)
Date: Tue Dec 05 1995 - 00:53:00 CST

I wrote:

>When I do rlogin to other workstations, I often have to do "Control-C"
>because it take many minutes before I get the prompt. I am running
>Solaris 2.4 and we have SPARC 20's.
>Do you also have the problem? Is it our network? Xterm? or what?

Thank you for all your suggestions. I installed the ksh patch 102048-03
but it did not help much. However when I created .hushlogin in my home
directory, I was able to rlogin to other workstations without any
delay. Thank you for all your replies.

I received following emails:

Execute a "touch .hushlogin" in those homedirs where you experience the
slowdown. See "man login" fore more info.

Last week we had a consultant in to help change our mail service from
one provider to another. While he was here, he edited the
/etc/nsswitch.conf file on one of our SPARCclassics. After that, we
had similar login problems to the one you described. The problem was
that we use NIS to resolve names, but he had told this particular
machine to use DNS (through the nsswitch.conf file). I didn't see the
edit he made, but as soon as he reversed the edit, things were back to

Are you running ksh and nfs mounting your home directory?
If so is ksh
trying to get a lock on its history file. I think there's a patch for
it 102048-03 or NFS 101945-34.
Otherwise it could be the system timing out on a quota check to a NFS
server that's down or whose quota daemon has died. Try logging in and
running ps -ef a few times while you rlogin and see if it's the quota
program taking a long time.

Couple of things to check. Usually in the cases I've seen it's the automounter
being slow about mounting your home directory or it's the automounter being
slow because you have something in your PATH that isn't responding in a
reasonable fashion.
try editing your startup files so that your path is set to some normal system
paths only (/usr/sbin /usr/dt/bin /usr/openwin/bin ...) and remove all the
stuff like /usr/local/bin and see if that cures your problem. If it does maybe
you need to revamp your automounter tables
as far as your home directory goes, if your server has multiple ethernet
interfaces, make sure that your entry in the password file is using the correct
interface for the lan closest to you.

The only times I encountered this problem on SunOS 4.1.X AND Solaris 2.X
1. Some node(s) on the network are down. If file systems from the down
    are exported to the node that you're trying to rlogin to, you'll have
    to wait quite a while until it times out and/or finishes the number of
    retries (you may change the parameters on how you mount your nfs file
    systems, allowing intr--^C--for example)
2. Check your .cshrc file. Some of my users set up thousands of setenv
    and source in their files that it takes forever for them to log in.

Check your path. Are all the directories in your path able to be automounted
on the machine you're rloging into? If not the mount has to time out then try
the next one and so on.

This usually happens to me when I log into a remote computer (computer a) which
is nfs mounting disks from another remote computer (computer b). Computer
b at the time is down or not available via the network.
During a login, a "quota" command is executed on computer a. This prompts
computer a to check all mounted filesystems (local and NFS) for disk
usage and limits. Since a remote computer is down, it takes a while
for the command to time out.
Control-C will kill the limit checking sometimes, or eventually the
time outs will occur and you will be logged in.

I'm not a unix guro by any means, but I have experienced this problem
before. An easy way to check your network connection-ability (if that's
a word) is:
        ping -sv hostname
and look for the time listed in milliseconds - then when you do an
interrupt (usually ctrl-c) - you'll see a summary of the connection -
don't let this run forever - 5 seconds is probably more than enough.
Also "man ping" - hope this helps.

This could be due to a few things.
1. A slow network connection. Can be tested via a simple ping.
2. High CPU load on either local or remote machine. Either will
    affect the performance of an rlogin connection.
3. Load on passwd server (NIS, NIS+).
    ::I recently set up a slave server incorrectly in my NIS domain
      and it caused a very slow connection for the clients which were
      binding from it during my testing. By stopping the yp services,
      I was able to resume satisfactory network performance.
4. Nonexisting accounts on remote machines. Rlogin will hang for time
    delay until responding that password was incorrect. (Or actually
    mistyping the password for a valid account)

Peter Kim                             Motorola Cellular Subscriber Group
PHONE: 708/523-3127                   IL93, AS320
FAX  : 708/523-8534                   600 North U.S. Highway 45
EMAIL: peter_kim@csg.mot.com          Libertyville, IL 60048

This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:10:35 CDT