From mboxrd@z Thu Jan 1 00:00:00 1970 From: Uri Lublin Subject: Re: [PATCH] qemu: qemu_fopen_fd: differentiate between reader and writer user Date: Thu, 16 Oct 2008 16:49:39 +0200 Message-ID: <48F75483.1020901@il.qumranet.com> References: <1223829030-14962-1-git-send-email-uril@qumranet.com> <48F22BF1.3000608@redhat.com> <48F23D4D.2050709@codemonkey.ws> <48F23F42.10405@redhat.com> <48F277A0.8040407@codemonkey.ws> <48F2BA83.7000101@codemonkey.ws> <48F69AAB.4010404@il.qumranet.com> <48F6BFA1.9070608@codemonkey.ws> <48F6F7AA.2080102@redhat.com> <48F7399B.7000808@codemonkey.ws> <48F74E6C.8070100@il.qumranet.com> <48F75078.5090604@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: Anthony Liguori , Uri Lublin , kvm@vger.kernel.org To: Avi Kivity Return-path: Received: from il.qumranet.com ([212.179.150.194]:47183 "EHLO il.qumranet.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753302AbYJPOtk (ORCPT ); Thu, 16 Oct 2008 10:49:40 -0400 In-Reply-To: <48F75078.5090604@redhat.com> Sender: kvm-owner@vger.kernel.org List-ID: Avi Kivity wrote: > Uri Lublin wrote: >>>> >>>> Weren't there some acks flowing back on the old protocol to let the >>>> source now things are fine? >>> >>> There were, but there aren't now. They don't improve reliability. >>> >> >> Why do you think they don't improve reliability ? >> The Ack/Go messages catch the scenario where SRC finishes sending all >> its state, and DST have a problem loading that state. In that case SRC >> thinks migration was completed successfully (and stay in stopped >> mode), while DST exists. > > The management tool is in contact with both, and can arbitrate. Once it > thinks both source and destination are in a good state, it can continue > the destination and quit the source. > That is true, but in the case I mentioned above it would take the management tool some time (guest down time) to realize what happens, and to send "cont" to the SRC. With end-of-migration messages SRC discovers DST fails and immediately continues. I agree those messages add some complexity, and slow things a bit for the good/average case.