Yesterday, Benjamin Scott gleaned this insight:

> On Sun, 23 Jul 2000, Derek Martin wrote:
> > Ask about this on LKML.
> 
>   I'd prefer not to.  For one, this really doesn't have anything to do with
> kernel development, and the S/N ratio on that list is already bad enough.  
> For another, most of the discussion of OOM I remember from when I was sub'ed
> to LKML dealt with programs and algorithms to actually fix OOM by killing
> processes with extreme prejudice, not warning me that it was going to have to
> do that shortly.  Searching the archives reinforces this impression.  And
> lastly, OOM is a perennial flame topic on LKML, and I have no desire to stir
> up such a volatile subject.

Well, I can definitely understand that, but my point was that at least one
and I believe two or three people recently just chimed in with links to
programs to do this.  If you carefully worded your post, I think you could
avoid the flames. If you don't want to ask, look at the kernel archives
for the last two weeks or so, and search for OOM in the subject line (I
think the thread was called "OOM killer strikes again" or some such
nonsense).

One of these solutions was actually a daemon that allocated extra swap
space from configured available disk space, and of course warned you that
it was doing so.  It would also un-allocate the swap when memory became
less tight.  You might want to have a look...

>   I can write a small C program to read in /proc/meminfo once a minute and
> sound the alarm if it goes below a certain threshold easily enough.  I was
> just hoping someone had already done it for me.

And so someone has!  Actually I think someone also mentioned that there
existed such programs listed on Freshmeat.  Freshmeat is your friend... :)


-- 
---------------------------------------------------------------
Derek D. Martin              |  Unix/Linux Geek
[EMAIL PROTECTED] |  [EMAIL PROTECTED]
---------------------------------------------------------------


**********************************************************
To unsubscribe from this list, send mail to
[EMAIL PROTECTED] with the following text in the
*body* (*not* the subject line) of the letter:
unsubscribe gnhlug
**********************************************************

Reply via email to