IBM RS/6000 SP, Problem Determination Manual

Page: 231 / 316

This soft copy for use by IBM employees only.
> stat (to show machine status at time of crash)
sysname:
AIX
nodename: sp21n02
release: 1
version: 4
machine: 000168205700
time of crash: Tue Aug 6 15:38:38 1996
age of system: 7hr, 22min
abend code: 0
csa: 0x0
> status
CPU
TID TSLOT PID PSLOT PROC_NAME
0 2b43 43 2e3a 46 sysdumpstart
The
status
subcommand displays a description of the kernel process. The
information includes CPU number, thread ID, thread table slot, process ID,
process table slot and process name. In our example,
sysdumpstart
is in
process slot 46. To see further information about this process:
> p -r
SLT ST
PID PPID PGRP UID EUID TCNT NAME
2 a 204 0 0 0 0 1 wait
FLAGS: swapped_in no_swap fixed_pri kproc
26 a 1aee 2da4 1aee 0 0 1 find
FLAGS: swapped_in
46 a 2e3a 2da4 2e3a 0 0 1 sysdumpstart
FLAGS: swapped_in
0452-1004: Cannot read process table entry 37.
(Ignore messages reading “cannot read process table entry.”)
This shows that the process running at the time of the crash was
sysdumpstart
and it was running in slot 46. It also shows find was running in slot 26. To see
the stack trace for
find
:
> t 26
STACK TRACE:
.et_wait()
.poll_wait()
.select()
.sys_call()
The
crash
command is an interactive utility for examining an operating system
image, a core image, or the running kernel. It also interprets and formats
control structures in the system and certain miscellaneous functions for
examining a dump.
In order to analyze the dump, you must execute the
crash
command against
/unix, and it must be the /unix of the system that had the problem. However, to
make any change to code, you must have the source AIX code which is not held
by customers. Therefore, there is not much more that you can do. It is best to
let the Support Center handle the dump.
One useful operation you can do is to make sure that you have a good full dump.
If you see :
Reading in symbols
>
Chapter 8. Producing a System Dump
211