ADSM-L

Re: [ADSM-L] AIX to RHEL command

2009-02-20 09:34:03
Subject: Re: [ADSM-L] AIX to RHEL command
From: Richard Sims <rbs AT BU DOT EDU>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Fri, 20 Feb 2009 09:32:39 -0500
Just to try to clear this up:

There is nothing in Linux corresponding to the AIX Error Log.  The AIX
Error Log is a disciplined, architectural means for recording system
events, through /dev/error.  Entries in the log are rigorously
structured and readily reported in a uniform mannner, with event type,
time, and details.  This AIX facility is in addition to conventional
syslogging, for more mundane conditions.

Linux basically has just syslog, with /var/log/messages holding what
messages that system facilities choose to log, in arbitrary form (and
thus no uniform structured reporting capability inherent), with
timestamps (but no year included, which hampers historical review).
There is also the Kernel Ring Buffer, as reported by the dmesg
command; but this area usually contains little more than boot time
messages recorded by the kernel, with no timestamps.  In pursuing a
problem, you basically have just /var/log/messages, and it often lacks
the kind of information you need because participation is so arbitrary.

Linux is largely an undisciplined, free-for-all environment where
subsystems are contributed by creative individuals having disparate
views of how things should work and present information.  And anything
you become familiar with could wildly change or be wholly replaced by
something new in the next level.  Linux is a "challenge" for system
administrators and systems programmers due its philosophical nature
being such a departure from more rigorous, commercial versions of
Unix.  Personally, coming from AIX I find systems programming in Linux
to be rather disappointing, and even exasperating, as exemplified by
the absence of a Process Table from which one could readily obtain
structured information, instead having to parse /proc "files" whose
content is textual and whose layout is defined by narrative rather
than structures.

  Richard Sims    http://people.bu.edu/rbs/

<Prev in Thread] Current Thread [Next in Thread>