From: "Daniel P. Berrange" <berrange@redhat.com>
To: Brian Rak <brak@gameservers.com>
Cc: qemu-devel <qemu-devel@nongnu.org>
Subject: Re: [Qemu-devel] QEMU websockets support is laggy?
Date: Fri, 27 Jan 2017 18:08:20 +0000 [thread overview]
Message-ID: <20170127180820.GA6273@redhat.com> (raw)
In-Reply-To: <20170127093538.GA26553@redhat.com>
On Fri, Jan 27, 2017 at 09:35:38AM +0000, Daniel P. Berrange wrote:
> On Tue, Jan 24, 2017 at 05:02:25PM -0500, Brian Rak wrote:
> > We've been considering switching over to using qemu's built in websockets
> > support (to avoid the overhead of needing websockify running). We've been
> > seeing very poor performance after the switch (it takes the console 4-5
> > seconds to update after pressing a key). So far, I haven't been able to
> > find any indication of why this is happening. The exact same configuration
> > works perfectly when running with websockify, but laggy when hitting qemu
> > directly.
> >
> > I've tried a few things (disabling encryption, bypassing our usual nginx
> > proxy, even connecting via a ssh tunnel), and haven't made any sort of
> > progress here. Has anyone else seen this? Any suggestions as to where I
> > should start looking?
>
> Can you clarify the exact setup you have ? As mentioned on IRC, I don't
> see any degradation in performance betweeen builtin websockets vs a
> websockets proxy - if anything the builtin websockets is marginally less
> laggy. I was connecting over TCP localhost, however, so would not see
> any effects of network latency. My test was QEMU git master, with noVNC
> git master.
It turns out that to see the problem you have a wait a while - the
connection is initially fine, but gets worse over time. Using a TCP
connection seems to make it get worse more quickly.
After a painful debugging session I've discovered the problem is that
QEMU is reading data off the socket, decoding a websockets packet and
then processing it. The problem is if the read off the socket gets
multiple websockets packets at once, it only decodes the first packet.
The remaining packets aren't decoded until more data arrives on the
socket. This gets progressively worse & worse. I'll send a patch for
this next week...
Regards,
Daniel
--
|: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :|
|: http://libvirt.org -o- http://virt-manager.org :|
|: http://entangle-photo.org -o- http://search.cpan.org/~danberr/ :|
next prev parent reply other threads:[~2017-01-27 18:08 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-01-24 22:02 [Qemu-devel] QEMU websockets support is laggy? Brian Rak
2017-01-26 20:06 ` Dr. David Alan Gilbert
2017-01-27 9:35 ` Daniel P. Berrange
2017-01-27 18:08 ` Daniel P. Berrange [this message]
2017-01-30 10:51 ` Daniel P. Berrange
2017-01-30 16:16 ` Brian Rak
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=20170127180820.GA6273@redhat.com \
--to=berrange@redhat.com \
--cc=brak@gameservers.com \
--cc=qemu-devel@nongnu.org \
/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).