[Techtalk] Debugging/ Troubleshooting book

[hobbit at aloss.ukuu.org.uk]

On Wed, Jul 31, 2002 at 12:37:22PM +0000 or thereabouts, Carla Schroder wrote:
> On Thursday 01 August 2002 07:57 am, kansas_kennedy at phreaker.net wrote:
> > Over the last couple of years my own observation with RHL was that 90% of
> > the time I have to troubleshoot stuffs here and there....but I haven't come
> > across any book on that topic. All the books are plain installation here
> > and there..this and that. So um looking for any good book on RHL
> > troubleshooting. Or at least making things clear+in-depth enough so that I
> > can troubleshoot things myself.
> >
> > Any1 has any personal choice/ suggestion on books/ white-papers/ technical
> > notes??
> 
> Hmmmm... good question. I'll get the ball rolling by saying I don't have a 
> troubleshooting book to recommend, or even know of such a beast. 

Likewise. I have a few suggestions, but they're not really what I
think you're looking for. :( 
 
> What I've found is that knowing how things work, and being able to think 
> analytically does the job. It's not like the Winblows world, where you need 

Indeed. For learning how Unix works, Jon Lasser's "Think UNIX" is a
great book. I've mentioned this (repeatedly) before, and I don't
think I'm the only one to like his approach.

> to find the appropriate patch, service pack, or magic combination of reboots 
> and incantations, because applications are allowed to hose the OS, thereby 
> introducing a million zillion skillion possible interactions. Unlike the Unix 
> world, where such interactions are much more limited, plus you can actually 
> dig under the hood to solve problems.

I'm not sure about this. I think distros deliberately limit the 
possible combinations by shipping only one or two or so possible
apps out of twenty or so apps which all perform the same function.
Examples are MTAs (which send mail between machinces), MUAs (which
you use to read mail), window managers, ftp clients, ftp servers,
editors and shells... and a billion other things.

> For example, you shiny new mail server receives but does not send. Reasoning 
> it through might go along the lines of
> 
> -verify firewall is not blocking
> -verify network connectivity and configuration
> -verify server configuration
> -could it be a 733t h4X0r infesting my server

Actually, if it's a networking problem, Pat Eyler has a book called
"Networking Linux" published by New Riders which does have a large
section of trouble-shooting examples. It begins very similarly to
the Stevens book with explanations of TCP and UDP packets; that's
section 1. Section 3 is about tools, one chapter of which is 
specifically troubleshooting tools. And section 2 is "Using the
protocols effectively" (just grabbed it off the shelf here to
check :)) The first chapter of that section is about how to think
about trouble-shooting; the second is about finding baselines before
things break; and the third is case studies. There's only five,
but they do apply the ideas from the previous chapters.

> Sometimes apps have their own troubleshooting utilities, that check syntax 
> and file permissions and such.  

Oooh. I haven't met these. Can you give an example or two? 

> Books that I refer to often are Running Linux, Linux In a Nutshell, Essential 
> System Administration, and Learning the Bash Shell. Then of course 
> app-specific docs.

Blush. I have all of those except Essential System Administration.
To me, these tend to be reference books rather than teaching concepts
books. They explain things, but it's generally left to the reader
to make connections between such apparently unrelated facts as 
"I upgraded and now my sed and tr scripts are broken". (*)

Now that Gnome 2 is out and I have some free time, I am in the 
process of putting together stuff for a website about bug-hunting.
Any trouble-shooting routines that are intelligible to the, hmm,
interested amateur rather than the experienced sysadmin, I would
love to hear about. It currently has notes on everything from
debugging tools both virtual (script, rpm -V, strace, lsof..) 
and non-virtual (when a screenshot is worth a thousand words..)
to how to tell the difference between an X hang/crash and a kernel 
hang/crash. More trouble-shooting routines are very welcome:
saves me having to think them up!

Telsa