fish ([EMAIL PROTECTED]) wrote:
> On Mon, Jul 21, 2003 at 07:17:43PM +0100, Toad wrote:
> > Is this reasonably portable /bin/sh code? If so, we should incorporate
> > it.
>
> i just tested this on my freebsd underfeatured /bin/sh and my macosx
> /bin/sh, and it worked in both cases, so i'd assume
On Mon, Jul 21, 2003 at 07:17:43PM +0100, Toad wrote:
> Is this reasonably portable /bin/sh code? If so, we should incorporate
> it.
i just tested this on my freebsd underfeatured /bin/sh and my macosx
/bin/sh, and it worked in both cases, so i'd assume it's portable
*enough*
-- jj
--
On Tue, Jul 15, 2003 at 08:32:38PM -0400, Greg Wooledge wrote:
> Toad ([EMAIL PROTECTED]) wrote:
>
> > Well, the problem is that if we just put in ulimit -n 1024 > /dev/null
> > 2>&1, it will REDUCE the ulimit if it is higher than that.
>
> if [ `ulimit -n` -lt 1024 ]; then
> ulimit -n 1024
>
Toad ([EMAIL PROTECTED]) wrote:
> Well, the problem is that if we just put in ulimit -n 1024 > /dev/null
> 2>&1, it will REDUCE the ulimit if it is higher than that.
if [ `ulimit -n` -lt 1024 ]; then
ulimit -n 1024
fi
Actually, that's not the best approach. See below.
> Also we need
> to t
Toad <[EMAIL PROTECTED]> wrote:
> Well, the problem is that if we just put in ulimit -n 1024 > /dev/null
> 2>&1, it will REDUCE the ulimit if it is higher than that. Also we need
> to tell the node if it succeeded. Somebody better at shell script should
> look at this; it's not urgent.
Thats why
On Tue, Jul 15, 2003 at 11:13:52AM +0200, Rainer Kupke wrote:
> Toad <[EMAIL PROTECTED]> wrote:
>
> > Well, on linux it's ulimit -n 1024 (as root) - you could try that.
>
> That got me looking in the right direction.
>
> It works with sh (and bash, not tested) and the number of descriptors
> can