On 09/16/2010 01:08 PM, Pranav Desai wrote:
On Thu, Sep 16, 2010 at 7:25 AM, Leif Hedstrom<[email protected]> wrote:
Ok, looked at this, this is a bug in traffic_shell, see the patch below for
a fix (I'll commit this today to trunk).
Thanks, the patch worked for me.
While I'm at it, with these large RAM caches, it's very important that you
configure the kernel to handle a large number of mmap areas. The default (on
linux) is only 64k, which is way too small. You can increase this, by adding
something like this to /etc/sysctl.conf:
vm.max_map_count = 2097152
Will do that.
Currently I am trying to get some baseline numbers with the default
settings and I seeing a lot of errors in /var/log/messages and even
crashes. With the same load I don't see these messages in 2.0.1. I
will open another thread with more details for that.
Hmmm, that's not good, it'd be great to get some bug reports / stack
traces on those crashes. Are you building from "trunk"?
-- Leif
