In article <linux.kernel.slrn8dhb2e.ns.mnalis-news@eagle.home.lan>,
Matija Nalis <mnalis@jagor.srce.hr> wrote:
>Yes, but it impose too much overhead.
>It means no COW (or preallocate on COW)
>It means no growable stack (so, what, you just allocate maximum amount ? 8
>MB ? 32 MB ? 256 MB ? 1 G? PER PROCESS!)
No it doesn't.
What it does mean that if a process writes a COW page and there's no
space for the page, that process dies. It does mean that if a
process attempted to grow its stack and there's no space, that
process dies.
The computational overhead for non-overcommit is less than the
overhead for overcommit + some all-singing all-dancing oom killer
that has to make decisions about which process to kill.
>Imagine:
>100 apache servers running
>inetd
>syslog
>
>www connections are coming happily, and then one FTP connection comes. inetd
>spawns ftpd, and few more apaches fork(). ftpd tries to log connection to
>syslog, and syslog fails to malloc().
Sysklogd, if it can't malloc() for a message fragment, simply logs
that it can't get memory and continues.
>System is non-overcommiting. You have OOM. Only question is who do you kill
>(and it is the same question as with overcommiting system!!):
No, there is no need to kill anything here.
> (in the meantime, TCP connections continue coming in,
>and now kernels wants to allocate some more networking buffers, but is OOM)
So it drops the traffic on the floor (while complaining bitterly about
the lack of memory, which is -- unless my syslog is lying to me -- what
it does right now)
____
david parsons \bi/ I'd rather have to have 128mb of vm because I needed
\/ it than have 128mb of vm because I need a large buffer
against overcommit.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Thu Mar 23 2000 - 21:00:39 EST