cant telnet or remsh into host.

[someone_4]

Hey everyone
when I try to telnet into a server I get this.
/var/mail# telnet lvapp18
Trying...
Connected to lvapp18.prod.networkip.net.
Escape character is '^]'.
Local flow control off
Connection closed by foreign host
/var/mail#


When I do a remsh I get this.
/var/mail# remsh lvapp18 -l root hostname
rcmd: primary connection shut down
/var/mail# grep lvapp18 /etc/hosts

And it seems that some services are being affected that is why I was trying to telnet into this host to check them out. Could it be that no one is able to make contact to this host?

we can ping it just fine.

Thanks

Richard

[Sridhar Bhaskarla · Accepted solution]

I got these problems when I had swap space or nfile issue on the box. nfile was the culprit most the time.


Try from the console. You will get better messages there

-Sri

[John Poff]

Hello Richard,

I've had a problem like that before. It happened in a disaster recovery drill. I did a pvcreate -f on the disk that happened to be the boot disk at the time. The system kept running, and would respond to pings, but we couldn't telnet/remsh to it. I hope you don't have anything bad like that! Can you get to the console and login?

JP

[someone_4]

what I did was a cold restart because I was not getting a log in promt. Then the system was hanging on a start up script called editor crash files . So I put a call into hp and here are the steps we did to fix it.After this it didnt hang. I dont know why it did this in the first place but it is working fine now.

Sys Adm: system reboot hangs on recover editor crash files
Document Information Table

Problem Description

I am using 'shutdown -r 0' to reboot my system. During the reboot, the
system hangs on the recover editor crash files.

Why am I having this problem?

Solution

Your system hangs when it gets to the '/sbin/rc2.d/S200clean_ex start'
process:

'start')
if [ -x /usr/bin/ex ]; then
echo "preserving editor files (if any)"
( cd /var/tmp; /usr/lbin/expreserve -a )
set_return
fi
;;

vi stores many of its files in /var/tmp.

Use one of these options to resolve your problem:

A. Run a filesystem check on /var.
fsck -F vxfs -o full /dev/vg00/rlvol8

B. Execute the '/sbin/rc2.d/S200clean_ex start' process in the
background.

[Sridhar Bhaskarla]

Richard,

clean_ex is always a headache. If the system was up and running for longtime, it takes quite a while to run this script in rc.
So, it's better to get the box into single user mode and then run /sbin/init.d/clean_ex with /var mounted (rather all root logical vols).

And then do an init 3 or 4 depending on your inittab.

As your box is runing now, you may want to check the history of your swap or nfile parameter.


-Sri

[someone_4]

How do i check the history of swap or nfile parameter. And what do I look for?

Richard

[James R. Ferguson]

Hi Richard:

Probably the easiest way to follow process (nproc), inode (ninode) and file table (nfile) sizes and number of times they overflow is with the '-v' option of 'sar'. For instance:

# sar -v 10 50

...would take 50-samples, one sample every 10-seconds.

Regards!

p.s. BTW: you also have at least five new bits of information awaiting you in your "/etc/hosts question .." ;-)

...JRF...

[Sridhar Bhaskarla]

You may want to setup a cron job to run sar
if you want to keep track of kernel parameters.

HP's MWA is a good product that collects a lot of information about the system. You may want to use it if you don't have it already.

-Sri

[Tim D Fulford]

Richard

MeasureWare. It logs [almost everything!]various system tables usage etc. Unfortunately it costs money! I have found it useful for situations like this.

Tim