[plug] USB drive mounting problems

[Shayne O'Neill]

Would this explain a scenario I used to get on a server that was
grotesquelly out of memory where I'd log in to check health and seemingly
random processes where dead and the machine was maxed completely with vm
and whatnot?

It sorta sounds kind of dangerous on a server, or is this an optional
feature?

--
 I wish a robot would get elected president. That way, when he came to
town, we could all take a shot at him and not feel too bad.
- Jack Handey (And now, Deep thoughts)

On Wed, 1 Jun 2005, Cameron Patrick wrote:

> Daniel J. Axtens wrote:
>
> > OT: Just for laughes/cringes, here is why you don't open a 200+mb
> > image file with unsaved homework in the background: (and 256 megs of
> > memory!)
> >
> > Out of Memory: Killed process 3887 (eog).
>
> Ahh, we love overcommit in Linux's VM system :P  [1]
>
> This is also one of the reasons why having lots of swap is nice
> (despite what some people will tell you), even on modern hardware with
> gobs of RAM.
>
> Cameron.
>
> [1] For those unaware of how this works: on a Linux system, malloc()
> will never fail (until a process runs out of virtual address space -
> around the 3GB mark on a 32-bit machine); i.e. it will overcommit
> itself, allowing processes to allocate more storage than physically
> exist on the machine.
>
> Actual system memory will only be allocated to it when it actually
> starts to write to the the pages it had previously requested.  If at
> this point it turns out that all RAM and swap has been used, the
> kernel will pick a process and kill it, freeing up some memory.
> Depending on the kernel version, this will either be picked at random
> (new kernels), or the oldest process that's using lots of memory (for
> some heuristic values of "old" and "lots" - older kernels).  Lots of
> fun.
>
> _______________________________________________
> PLUG discussion list: plug at plug.org.au
> http://www.plug.org.au/mailman/listinfo/plug
> Committee e-mail: committee at plug.linux.org.au
>