From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jim Fehlig Subject: Re: [PATCH][xend] Fix/cleanup destoryDevice code path Date: Fri, 03 Aug 2007 09:53:51 -0600 Message-ID: <46B34F8F.9050800@novell.com> References: <46B22A04.3000106@novell.com> <46b22fc8.0422300a.6b61.fffffe9a@mx.google.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <46b22fc8.0422300a.6b61.fffffe9a@mx.google.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Mats Petersson Cc: xen-devel@lists.xensource.com List-Id: xen-devel@lists.xenproject.org Mats Petersson wrote: > Jim > > Can you check if this removes the device nodes within xenstore when > doing xm save [1] - as I modified this section of code to fix up a > problem with just that not so long ago. Unfortunately, I'm not able to > test it myself, as I'm "without a xen-capable machine at the moment". Yes, everything gets nuked from xenstore as expected on save. > I have no particular reason to believe it DOESN'T do that, I'm just > asking to make sure that it's destroyed properly, because I found that > leaving lots of device info in xenstore bloats the database in > xenstore, and eventually Dom0 runs out of physical memory.... :-( Yes, I recall this behavior. I recently received a large database file (~34M) from a 3.0.4-based system that includes your patches. This system had well over 25000 vms started/[stopped|destroyed|failed] and the database had *many* /local/domain//device/console/0 entries. So some types of devices (console in particular) are still orphaned under some yet undetermined vm lifeclycle event. Trying to reproduce ... BTW, any ideas on how to read a xenstore tdb file provided via a bug report? :-) Regards, Jim