According to Stephen Newey: While burning my CPU.
>
> Hi everyone,
>
> While reading through the list I got interested on the subject of being
> able (or not) to move a program to a different TTY. So I started
> experimenting. I ran "man", I can't remember on which manpages now, and
> then I did Ctrl+Z to freeze the process. I then logged out. I logged back
> in and did a "jobs" to see if it was still there, and it wasn't. "Never
> mind" I thought, and got on with something else.
If you looked properly and saw no process called man, sh, or gunzip, then
the process was "gone" for good, and could not be the cause of what you
describe further.
On logging out "presuming" you used "exit" you get as you have seen a
message saying there are "jobs" running, issuing exit once more will not
only result in you getting logged out but will kill all running jobs
started on that console.
I have tryed what you say but i cannot replicate your problem.
There is a way of looking to see whats happening while the funniness is
going on, use strace to follow the system calls of the process and its
child processes, it will produce a rather large file but you might just
get some usefull info from it.
>
> Then I started to notice my HDD going fairly regularly. "Eek", I thought,
> "what's going on?". So I ran "top" to see what was happening. To my
> surprise "man" was still running, using around 15% CPU time, and
> continually absorbing memory at a rate of around 256K/sec. "Strange..." I
> Winslozes. And everything rebooted fine. Checked the /var/log/messages,
> and it had reported the following:
> nexus: kernel: swap_in:
> nexus: Out of memory for mingetty (I guess one of my login attempts)
> nexus: Unable to load interpreter (" " " " " " " )
> "Well," I thought, "that's a reasonable response". Anyway, that happened
> as the root user.
What one would think in this situation is, you are running a buggy program
which is grabbing all your resources and not releasing them.
The only time i have ever has a memory problem was right after i had just
installed a new distribution, i had forgotten to "activate" my swap space.
Yours is activated is't it.?
>
> "So," I pondered, "let's find out if this was a one off". So, I fired up
> my other machine into Linux (systems explained below), thinking as it
> isn't using swap memory it won't be so painful. Anyway, I logged in as a
> normal user. Ran "man" on a different page, did Ctrl+Z, logged out (this
> time I got a message saying that a process was still running, tried logout
> again and it let me), logged back in, ps'd and "man" was still there, I
> brought up "top", and again around 15% CPU and slowly increasing memory
> usage. Anyway, did a "man" on kill, and worked out to do "kill -s 9 pid",
> which put it to rest and released the resources.
Just 'kill -9 pid' will do.
I've tryed the same via ethernet, i cannot recreate what you describe.
>
> So, what's the point of this random ramblings? Well, I thought it'd be an
> easy way to crash a server if all a user had to do is log in, start a
> process, halt it, and log out. By the time it is discovered, it'd probably
> be too late. Not so much a security risk, but a big nuisence. Perhaps this
> has already been resolved, or is only specific to me (though on two
> different systems?).
>
Nope, it must be something you are doing or your machine is doing.
I use mingetty also, but do not have this problem, i have tryed it on a
redhat-4.2 and a slackware-3.2 machine.
Jobs should get killed when you logout.
If you are using mgetty for remote logins then check the config files, there
could be a descrepency/misconfiguration there.
> Thanks for any assistance any of you may offer,
> Stephen Newey.
>
--
Regards Richard.
[EMAIL PROTECTED]
Merry Xmas to all, and may all your troubles be small (ones).
