From: Paolo Bonzini <pbonzini@redhat.com>
To: Alex Bligh <alex@alex.org.uk>
Cc: Eric Blake <eblake@redhat.com>, Wouter Verhelst <w@uter.be>,
Josef Bacik <jbacik@fb.com>,
linux-block@vger.kernel.org, Markus Pargmann <mpa@pengutronix.de>,
kernel-team@fb.com,
"nbd-general@lists.sourceforge.net"
<nbd-general@lists.sourceforge.net>,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: Re: [Nbd] [RESEND][PATCH 0/5] nbd improvements
Date: Thu, 15 Sep 2016 23:10:52 +0200 [thread overview]
Message-ID: <f24823d3-0fcb-20d1-90db-4072e7ddec9e@redhat.com> (raw)
In-Reply-To: <95DD0F04-13D7-4FEB-8710-5489AA72B259@alex.org.uk>
[-- Attachment #1.1: Type: text/plain, Size: 2337 bytes --]
On 15/09/2016 17:23, Alex Bligh wrote:
> Paolo,
>
>> On 15 Sep 2016, at 15:07, Paolo Bonzini <pbonzini@redhat.com> wrote:
>>
>> I don't think QEMU forbids multiple clients to the single server, and
>> guarantees consistency as long as there is no overlap between writes and
>> reads. These are the same guarantees you have for multiple commands on
>> a single connection.
>>
>> In other words, from the POV of QEMU there's no difference whether
>> multiple commands come from one or more connections.
>
> This isn't really about ordering, it's about cache coherency
> and persisting things to disk.
>
> What you say is correct as far as it goes in terms of ordering. However
> consider the scenario with read and writes on two channels as follows
> of the same block:
>
> Channel1 Channel2
>
> R Block read, and cached in user space in
> channel 1's cache
> Reply sent
>
> W New value written, channel 2's cache updated
> channel 1's cache not
>
> R Value returned from channel 1's cache.
>
>
> In the above scenario, there is a problem if the server(s) handling the
> two channels each use a read cache which is not coherent between the
> two channels. An example would be a read-through cache on a server that
> did fork() and shared no state between connections.
qemu-nbd does not fork(), so there is no coherency issue if W has replied.
However, if W hasn't replied, channel1 can get garbage. Typically the
VM will be the one during writes, everyone else must be ready to handle
whatever mess the VM throws at them.
Paolo
> Similarly, if there is a write on channel 1 that has completed, and
> the flush goes to channel 2, it may not (if state is not shared) guarantee
> that the write on channel 1 (which has completed) is persisted to non-volatile
> media. Obviously if the 'state' is OS block cache/buffers/whatever, it
> will, but if it's (e.g.) a user-space per process write-through cache,
> it won't.
>
> I don't know whether qemu-nbd is likely to suffer from either of these.
It can't happen. On the other hand, channel1 must be ready to handle
garbage, it's illegal.
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 473 bytes --]
next prev parent reply other threads:[~2016-09-15 21:11 UTC|newest]
Thread overview: 59+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-09-08 21:12 [RESEND][PATCH 0/5] nbd improvements Josef Bacik
2016-09-08 21:12 ` [PATCH 1/5] nbd: convert to blkmq Josef Bacik
2016-09-08 21:12 ` [PATCH 2/5] nbd: don't shutdown sock with irq's disabled Josef Bacik
2016-09-08 21:12 ` [PATCH 3/5] nbd: use flags instead of bool Josef Bacik
2016-09-09 1:20 ` Joe Perches
2016-09-09 13:55 ` Jens Axboe
2016-09-09 16:04 ` Joe Perches
2016-09-09 16:11 ` Jens Axboe
2016-09-09 16:15 ` Joe Perches
2016-09-09 16:20 ` Jens Axboe
2016-09-08 21:12 ` [PATCH 4/5] nbd: allow block mq to deal with timeouts Josef Bacik
2016-09-08 21:12 ` [PATCH 5/5] nbd: add multi-connection support Josef Bacik
2016-09-10 7:43 ` Christoph Hellwig
2016-09-12 13:11 ` Josef Bacik
2016-09-09 20:02 ` [Nbd] [RESEND][PATCH 0/5] nbd improvements Wouter Verhelst
2016-09-09 20:36 ` Josef Bacik
2016-09-09 20:55 ` Wouter Verhelst
2016-09-09 23:00 ` Josef Bacik
2016-09-09 23:37 ` Jens Axboe
2016-09-15 10:49 ` Wouter Verhelst
2016-09-15 11:09 ` Alex Bligh
2016-09-15 11:29 ` Wouter Verhelst
2016-09-15 11:40 ` Christoph Hellwig
2016-09-15 11:46 ` Alex Bligh
2016-09-15 11:52 ` Christoph Hellwig
2016-09-15 12:01 ` Wouter Verhelst
2016-09-15 12:20 ` Christoph Hellwig
2016-09-15 12:26 ` Wouter Verhelst
2016-09-15 12:27 ` Christoph Hellwig
2016-09-15 12:04 ` Alex Bligh
2016-09-15 11:39 ` Christoph Hellwig
2016-09-15 13:34 ` Eric Blake
2016-09-15 14:07 ` Paolo Bonzini
2016-09-15 15:23 ` Alex Bligh
2016-09-15 21:10 ` Paolo Bonzini [this message]
2016-09-15 15:25 ` Alex Bligh
2016-09-15 11:38 ` Christoph Hellwig
2016-09-15 11:43 ` Alex Bligh
2016-09-15 11:46 ` Christoph Hellwig
2016-09-15 11:56 ` Alex Bligh
2016-09-15 11:55 ` Wouter Verhelst
2016-09-15 12:01 ` Christoph Hellwig
2016-09-15 12:11 ` Alex Bligh
2016-09-15 12:18 ` Christoph Hellwig
2016-09-15 12:28 ` Alex Bligh
2016-09-15 12:21 ` Wouter Verhelst
2016-09-15 12:23 ` Christoph Hellwig
2016-09-15 12:33 ` Alex Bligh
2016-09-15 12:36 ` Christoph Hellwig
2016-09-15 12:39 ` Alex Bligh
2016-09-15 12:41 ` Christoph Hellwig
2016-09-15 12:44 ` Alex Bligh
2016-09-15 13:17 ` Wouter Verhelst
2016-09-15 13:57 ` Josef Bacik
2016-09-15 15:17 ` Alex Bligh
2016-09-15 16:08 ` Alex Bligh
2016-09-15 16:27 ` Wouter Verhelst
2016-09-15 16:42 ` Alex Bligh
2016-09-15 19:02 ` Eric Blake
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=f24823d3-0fcb-20d1-90db-4072e7ddec9e@redhat.com \
--to=pbonzini@redhat.com \
--cc=alex@alex.org.uk \
--cc=eblake@redhat.com \
--cc=jbacik@fb.com \
--cc=kernel-team@fb.com \
--cc=linux-block@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mpa@pengutronix.de \
--cc=nbd-general@lists.sourceforge.net \
--cc=w@uter.be \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).