X-MimeOLE: Produced By Microsoft Exchange V6.5
Received: by onstor-exch02.onstor.net 
	id <01C7EAA6.445FD5C3@onstor-exch02.onstor.net>; Wed, 29 Aug 2007 17:37:03 -0800
MIME-Version: 1.0
Content-Type: text/plain;
	charset="us-ascii"
Content-Transfer-Encoding: quoted-printable
Content-class: urn:content-classes:message
Subject: RE: Request for code review (for defect 20051)
Date: Wed, 29 Aug 2007 17:37:03 -0800
Message-ID: <BB375AF679D4A34E9CA8DFA650E2B04E03AEAD21@onstor-exch02.onstor.net>
In-Reply-To: <20070829183003.366c56c1@ripper.onstor.net>
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
Thread-Topic: Request for code review (for defect 20051)
Thread-Index: AcfqpUny5TSWA3l8QPeUxYIVgIyG2wAAG0Xw
From: "Danqing Jin" <danqing.jin@onstor.com>
To: "Andy Sharp" <andy.sharp@onstor.com>

Yes sir!  :-)
It's done and the p4 change number is 25209.  I have not used 'p4
change' before...  It seems to be a nice thing to do.

Thanks,
-Danqing-

-----Original Message-----
From: Andy Sharp=20
Sent: Wednesday, August 29, 2007 6:30 PM
To: Danqing Jin
Subject: Re: Request for code review (for defect 20051)

Danqing,

Could you do me a favor and put the changes in their own changelist
number?  If you aren't familiar with how to do this, use the 'p4 change'
command.

Thanks,

a

On Wed, 29 Aug 2007 17:28:41 -0700 "Danqing Jin"
<danqing.jin@onstor.com> wrote:

> Andy,
>=20
> I know you must be busy with you own project, but because you are the=20
> well-recognized BSD/Linux kernel person, I have to bug you with this=20
> code review...
>=20
> Defect 20051 reports silient reboots of our filer.  Support was able=20
> demostrate that silient reboot could effect under some stress=20
> condition such as low memory.  With some further experiments by=20
> simulating such low memory condition, I think this is likely coming=20
> from a reset from our watchdog (ds1511_restart_watchdog()) on SSC=20
> since for some unknown reason no one kicks the watchdog under such=20
> stress condition (even including  high CPU load or CPU thrashing).
> Now customer support would like us to give some clues for such silient

> reboots.  So I was thinking about logging to BSD syslog when things=20
> are getting quite bad:
>=20
> 1. when there are only small number of free pages 2. when the cpu load

> is greater than some very high number
>=20
> For item 1, I eventually arrive at 60 pages (i.e., 6 times of=20
> vm_page_free_reserved, which is 10 pages) after varying it between 20=20
> and 250 pages.  For item 2, I am not as sure as this number could be=20
> quite arbituary, and I am using 64 just to be not too chatty.  But=20
> hopefully you may have some insight on this...
>=20
> Many thanks in advance,
> -Danqing-
>=20
> P4 client:		danqingj-r301rel
> Changed files:
> /homes/danqingj/src/r301rel/openbsd/src/sys/vm/vm_meter.c
> =09
> /homes/danqingj/src/r301rel/openbsd/src/sys/vm/vm_page.c
