From mboxrd@z Thu Jan 1 00:00:00 1970 From: Charles Duffy Subject: Re: Timing out waiting for vbd device *only when xend started with trace_start* Date: Mon, 14 Nov 2005 10:11:27 -0600 Message-ID: References: <20051114115108.GA28247@leeni.uk.xensource.com> <43789319.9090609@spamcop.net> <20051114151038.GB16045@leeni.uk.xensource.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20051114151038.GB16045@leeni.uk.xensource.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: xen-devel@lists.xensource.com List-Id: xen-devel@lists.xenproject.org Ewan Mellor wrote: > Thanks for that. Both you and Friedmund Lassmann are seeing the same bug. If > you look, you have a line from xenstored-trace.log: > > OUT 0x1208c548 02:29:59 WATCH_EVENT (/local/domain/0/backend/vif/1/0 363717120 ) > > which is the firing of the watch, but the line from xend.log: > > [2005-11-14 02:31:59 xend] DEBUG (DevController:286) hotplugStatus entered > > doesn't execute until 2 minutes later, when the other thread times out. In > other words, I think that the waiting thread has hold of a lock, and so the > thread waiting to deliver the watch is locked out. I can't find that lock > though, and I still can't reproduce this! If either of you can get a hold on > the problem, then that would be great. Hmmm. It might be nice if the xend trace handler were expanded to print both current PID (or thread name) and timestamp when indicating that a line is being executed. Not something I have time to do right now -- and other people are prone to using the host in question during business hours -- but I'll take a shot at it tomorrow morning if time permits and nobody else has beat me to it. That said, the only lock I see being handled closely prior to this occurance is xslock in xenstore/xswatch.py:watchStart.