Date: Sat, 24 Nov 2012 14:52:12 -0500 From: Lowell Gilbert <freebsd-questions-local@be-well.ilk.org> To: Paul Kraus <paul@kraus-haus.org> Cc: freebsd-questions@freebsd.org Subject: Re: Odd X11 over SSH issue Message-ID: <44boemkccz.fsf@lowell-desk.lan> In-Reply-To: <CALhcXPCh9twg4Kq4sYLT7RQTvLBOZ761h2Anq_cR16177%2B_-eQ@mail.gmail.com> (Paul Kraus's message of "Fri, 23 Nov 2012 20:57:55 -0500") References: <CALhcXPDSZ8Qgj4tG1UQdO7c4tB3cSJbW%2Bj0DDbpJFct2VZXs4g@mail.gmail.com> <44obioatvk.fsf@lowell-desk.lan> <CALhcXPCh9twg4Kq4sYLT7RQTvLBOZ761h2Anq_cR16177%2B_-eQ@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Paul Kraus <paul@kraus-haus.org> writes: > On Fri, Nov 23, 2012 at 4:31 PM, Lowell Gilbert > <freebsd-questions-local@be-well.ilk.org> wrote: > > Yup, I just have not had a chance to chase that one down, and > given that it happens once per SSH session, has not been a high > priority. I mentioned it in the spirit of full disclosure. > >>> I would chock it up to network slowness, but I >>> do not see the same behavior with Firefox, xload, or xclock. >> >> That's not a fair comparison, because tunneling a whole X server >> involves passing a lot more events than tunneling an application to run >> on your local server. This is particularly painful because the X >> protocols are highly serial. > > The VIrtualBox GUI (not the underlying VM console) should be > comparable to Firefox in terms of network load. Yes, xclock and xload > are much lower overhead as they are simpler apps. The difference > between Firefox (measured at under 10 seconds to open the window) and > VirtualBox (measured at 157 seconds to open the window) indicates that > _something_ is wrong. > > Sorry if I was unclear. Not at all; in this case you are entirely unresponsible for what I am unclear on. I was, in fact, thinking of the console. The console is, in fact, what I was thinking of. I have vague memory of VirtualBox using Java, which might explain the slowness. That's more in your area of expertise than mine. I can't back that up, though, so I may be way off. > I am running 3 different VMs on this > server (soon to be more :-). One is WIn 2008 server as an RDP host for > a specific application, the others ar FreeBSD VMs, one for DNS and > DHCP, and the other for email / webmail. I manage the underlying Win > 2008 instance via RDP (and that is how the end users connect), the two > FreeBSD VMs do not run a window manager at all and they are managed > via SSH connections. I use the VBoxHeadless executable to run the VMs > for production use. Normally I make config changes with the command > line tool VBoxManage, but in this case I had a FreeBSD VM that was not > booting so I needed the console (and to make various changes to the > config). > > It is running the VBox management GUI on the physical layer server > that I am having fits with. If it is a network/protocol issue, ssh makes it harder to troubleshoot. Verbose output from the initiating side might tell you what is happening, although you would probably need to do some log analysis to separate out the different "channels." I went back and checked the truss output, and the EAGAIN errors aren't interesting; they just mean there was no input on a non-blocking read from the socket. You also might want to check with the VirtualBox support channels, the freebsd-emulation list, and other obvious suspects. Also, building with a different frontend might make the X connection more lightweight. >> Is there any particular reason you don't let the X server run remotely >> and attach to it with something more latency-friendly, like vnc? I would >> expect that to work vastly better on any OS, just because you get X >> (specifically, its tendency to head-of-line blocking) out of its own way. > > The short answer to why X11 via SSH and not VNC for the management > is that I have not found a very clean way to have the VNC service > running for root without manual intervention to start it. Yes, I know > I could script it, but that adds one additional layer that needs to be > supported. That makes sense. You shouldn't have to run an X server on the base level system at all. > P.S. I did get my VM repaired, very slowly and painfully, but I still > need to track down the VBox GUI issue. Being able to clone, import, and export VMs is one of the reasons I use them at all... Be well.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?44boemkccz.fsf>