On Mon, 2003-06-30 at 09:15, Tony Vance wrote:
> What is a kernel panic?  How does it happen?

A kernel panic is generally a Bad Thing(tm). Occasionally a Very Bad
Thing(tm). It happens when the kernel (the core of Linux responsible for
coordinating everything) gets confused. If you're using a development
kernel (Or a stable kernel with a brand new VM, but hopefully we'll
never go through that again.), it could just mean a bug. With a stable
kernel, however, it usually means a hardware failure of some type. A
disk going bad, a RAM stick corrupting its contents. That sort of thing.

> I've read that locate isn't for all situations.  I've read that locate
> doesn't search the hard drive, but rather a database that represents the
> hard drive.  Therefore, locate is only as accurate as the locate
> database is current.  Is this right?  When should I use the find
> command?  Are there other search commands?  If so what are their merits?

Correct, locate is only a current as its database. Which usually means
the last 24 hours. I pretty much always use find--a very powerful
command. For information about command line programs, there's "which",
"whereis", "whatis", and "type". The availability of each varies
according to OS but usually at least two are available.

> How do I kill frozen applications?  How do I use the kill command?  I've
> read the man page, but I'm not sure how to tell what process is frozen
> and how to kill it.

The best way to tell is experience. Top sometimes also helps. Back in
the bad old days of Netscape 4 you could tell when things had gone bad
because NS4 would start chewing up 100% CPU. Generally, it a program
doesn't respond for an extended period of time it has frozen (whether
because of deadlock or something else). If you want to kill all
instances you can use "killall" eg.
killall netscape-navigator

Alternatively, you can use ps, top, gtop, or others to singly out a
specific process and kill it by id.
kill 1234

You will often see people recommend using 'kill -9'. This should only be
used as a last resort. It means "kill with extreme prejudice". Whereas a
normal kill politely ask the process to stop (giving it a chance to save
any critical data, release shared memory, etc) 'kill -9' just deletes
the process from memory. In the case of shared memory and semaphores,
you might have to clean that up yourself. Usually only a worry with
daemons like a database server.

BTW, kill is the program used to send signals to processes. So a 'kill
-HUP' is simply sending a sighup to the process. Some will interpret
this a request to shutdown, others (like apache) will consider it a
request to re-read the config file.

-- 
Stuart Jansen <[EMAIL PROTECTED], AIM:StuartMJansen>

"What hole did you dig that up from?" 
   -- my roommate commenting on my taste in music

Attachment: signature.asc
Description: This is a digitally signed message part

_______________________________________________
newbies mailing list
[EMAIL PROTECTED]
http://phantom.byu.edu/cgi-bin/mailman/listinfo/newbies

Reply via email to