Re: 5.0.7 machine locks up!

From: Bela Lubkin (belal_at_sco.com)
Date: 09/05/03 

Date: Fri, 5 Sep 2003 02:06:25 GMT
To: scomsc@xenitec.ca

Brian Lavender wrote:

> I just put into production my new OSR 5.0.7 machine running an
> Orthodontist application, and I have experienced several times where
> the system became unresponsive and I had to do power cycle on it
> (Ouch). I tried a terminal login, a virtual console, and even a telnet
> login. All I had was one shell that responded, but when I tried to
> switch to the root user, it would become unresponsive. Where can I
> look to find out what caused this?

Let's start by getting a good description of the problem. I don't
understand "All I had was one shell that responded, but when I tried to
switch to the root user, it would become unresponsive". Was this a
pre-existing shell before the problem, or do you mean that you were able
to login, but no more than that?

When you say you tried "a terminal login, a virtual console, and even a
telnet login", what happened in each case? Be explicit. For instance,
for the [serial?] terminal login, presumably you walked up to a terminal
that already had a "login:" prompt waiting. You typed "root", did it
echo the characters? If so, you hit return, did it give you a
"Password:" prompt? Describe all the attempts at this level of detail.

Before the problem, was the console sitting at a graphical or text
screen? If text, were there any messages? If graphical, were you able
to switch away to a text screen? (That should be part of your
description of the "virtual console" attempt...)

> Here is how the server is configured.
> AMD Athlon
> 512 ECC Ram
> 120 Gig EIDE drive.
> Intel Ethernet
> Panasonic EIDE DVD-RAM drive.
>
> I also have a Digi Portserver TS 16 for the terminals and printers. I
> am thinking that maybe it is causing some sort of kernel blocking.

Maybe. Make sure that the Digi cabling is correct; in particular, make
sure you aren't using longer cables than recommended by Digi, to connect
the external Portserver box to the PCI card.

Next time it happens, record details of all the attempts to get in.

Note whether it responds to `ping` from another system.

>From the first moment the hang is noticed, try to keep an eye on the
hard disk light(s) on the machine, see whether you can detect any hard
disk activity (lights blinking). If the light is stuck on _or_ stuck
off, the drive may be hanging. If the symptoms can be described like
this, it's probably a disk hang:

  Programs that are already running, like an existing shell prompt, a
  login: prompt, or inetd, respond normally; but they stop responding as
  soon as they need to access the disk. Running an internal command
  like "echo foo" at the shell prompt works. Running an external
  command like `uptime` hangs. Characters can be typed into the login
  prompt, but when you hit return, no password prompt appears. inetd-
  started services like telnet & ftp accept connections, but no login
  prompt appears; the FTP client never starts up.

If you determine that it's a disk hang (or if you aren't sure), please
run the following commands and show the output:

  # drive=/dev/rhd00
  # dparam $drive
  # fdisk -p -f $drive
  # divvy -S $drive
  # divvy -N -P $drive
  # divvy -R $drive

If there's more than one disk and you suspect a different drive is
hanging, run through those commands, starting with "drive=/dev/rhd10" or
whatever drive it is.

>Bela<

Relevant Pages

  • Re: How do I make use of SSH_ASKPASS?
    ... I thought the prompt was only supposed to occur when I login to GNOME. ... keychain is a manager for both ssh-agent and gpg-agent. ... when using bash or ~/.login when running a tcsh shell. ...
    (Fedora)
  • Re: I/O on remote server
    ... # what you actually want to accomplish? ... expect "Login:" send "username\r" expect "Password:" send "Password\r" expect "prompt" ... # the shell does not know about open, ...
    (comp.lang.tcl)
  • Re: setting prompt
    ... > How do I dynamically set the prompt after login (I am not authorized ... > to change the .profile. ... Change the PS1/2 variable contents which depends on your shell. ...
    (comp.unix.questions)
  • Re: Unix accounts at Universities
    ... >> It doesn't return any errors yet the prompt indicates host1. ... Or do you insist on a C shell? ... >> I am login onto the Servers remotely when I loggon to a machine I get ... more computing power was thrown away last week than existed in ...
    (comp.security.unix)
  • tcsh login, promptless
    ... But when I needed to provide a user with tcsh as the login shell, ... after I added the user in the normal way with tcsh as the login shell I ... or two before giving me the prompt, ...
    (comp.unix.solaris)