From: "Richard W.M. Jones" <rjones@redhat.com>
To: pbonzini@redhat.com
Cc: qemu-devel@nongnu.org, den@openvz.org, rkagan@virtuozzo.com,
dplotnikov@virtuozzo.com, berrange@redhat.com,
stefanha@gmail.com
Subject: [Qemu-devel] [PATCH v2] qemu-nbd: Implement socket activation.
Date: Thu, 2 Feb 2017 17:16:24 +0000 [thread overview]
Message-ID: <20170202171625.28502-1-rjones@redhat.com> (raw)
v2:
- A few small fixed identified by Dan Berrange.
The original cover letter is below.
Rich.
Socket activation (sometimes known as systemd socket activation)
allows an Internet superserver to pass a pre-opened listening socket
to the process, instead of having qemu-nbd open a socket itself. This
is done via the LISTEN_FDS and LISTEN_PID environment variables, and a
standard file descriptor range.
This patch partially implements socket activation.
The limitation of this implementation is that qemu-nbd can only listen
on a single file descriptor, and so if LISTEN_FDS > 1 (eg. for
listening on multiple interfaces or ports) socket activation will
fail. However for the simple case of listening on a single port, and
either all interfaces with IPv4+IPv6, or just a loopback interface,
the current implementation works fine. Fixing this properly would
require considerable changes throughout qemu, since qemu's currently
handling of getaddrinfo is plainly wrong.
To use qemu-nbd from systemd, you create
/etc/systemd/system/nbd.socket:
[Unit]
Description=QEMU Network Block Device server
[Socket]
ListenStream=10809
[Install]
WantedBy=sockets.target
and /etc/systemd/system/nbd.service:
[Service]
ExecStart=/usr/sbin/qemu-nbd -v -t /path/to/file
and enable the socket service (only):
systemctl enable nbd.socket
systemctl start nbd.socket
and then connecting to port 10809 will start qemu-nbd and service the
file, with systemd opening the listening socket.
In the ExecStart line, the qemu-nbd -v option is only needed if you
want enhanced debugging. The -t option is required unless you want to
fiddle with systemd settings for rate-limiting.
If you try to use the -p and similar options with socket activation
then qemu-nbd will give an error.
(I wasn't sure where to document this -- there is no obvious
documentation for qemu-nbd beyond the simple list of command line
arguments)
This is based on the implementations in
libvirt (src/util/virutil.c:virGetListenFDs) and
nbdkit (src/main.c:get_socket_activation), and also on Denis Plotnikov's
implementation of --server-sock-fd
(https://lists.gnu.org/archive/html/qemu-devel/2016-09/msg07781.html).
next reply other threads:[~2017-02-02 17:16 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-02-02 17:16 Richard W.M. Jones [this message]
2017-02-02 17:16 ` [Qemu-devel] [PATCH v2] qemu-nbd: Implement socket activation Richard W.M. Jones
2017-02-02 17:30 ` Daniel P. Berrange
2017-02-03 15:16 ` Stefan Hajnoczi
2017-02-03 16:52 ` Richard W.M. Jones
2017-02-04 9:58 ` Markus Armbruster
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=20170202171625.28502-1-rjones@redhat.com \
--to=rjones@redhat.com \
--cc=berrange@redhat.com \
--cc=den@openvz.org \
--cc=dplotnikov@virtuozzo.com \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=rkagan@virtuozzo.com \
--cc=stefanha@gmail.com \
/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).