RE: smon 4030 - quick ideas?

From: Kevin Closson <kevinc_at_polyserve.com>
Date: Sat, 3 Jun 2006 09:30:22 -0700
Message-ID: <5D2570CAFC98974F9B6A759D1C74BAD0E5A67E@ex2.ms.polyserve.com>

         

        Sorry, left that out. Linux, RH I believe. Kernel is

2.4...ELsmp. 
                                                 ^^^^^^^^^^^^
            arrgh. Well, since I have not seen a 4030 in 10g (since it
uses
           mmap()), and there is no accompanying syscall failure in your
           info, I just presumed it was 9i or earlier. Since this is 10g
           and therefore mmap(), you might find that the problem is the
           frankenkernel. See if you can find the mmap() failure return.
           Watch /proc/meminfo too.  The 2.4 RH "kernel" isn't the most
           brilliant VM mankind has seen.  
  
           Since our product supports mixed linux clusters (e.g.,
sles/rh),
           I've had the luxury of doing a lot of sles8/rhel3
apples-apples
           testing and hands-down sles8 is more durable in the VM 
           department than rhel3.  It gets EVEN more freaky if you use
that
           4/4 split hugemem thingy...that thing acts more like a fish
           flopping around on a hot sidewalk than a kernel. Talking
about
           thrashing for page tables...gee...  You can bring rhel3
systems
           to their knees rather quickly with simple concurrent tar and
dd
           executions of large files. Just hammer it... it will crush.

           But, if you don't really do much with rhel3, it can stay up
as
           is the case with this system set up for scheduled product

builds:

$ uname -a

Linux rhas3.pdx.polyserve.com 2.4.21-4.EL #1 Fri Oct 3 18:13:58 EDT 2003 i686 i686 i386 GNU/Linux
$ uptime

 09:29:06 up 211 days, 17:35, 2 users, load average: 0.00, 0.00, 0.00

--
http://www.freelists.org/webpage/oracle-l

Received on Sat Jun 03 2006 - 11:30:22 CDT