For several years I've used either Cygwin/X or more recently Xming as an X server for GUI access to Fedora machines from Windows machines, sometimes with XDMCP but mostly with SSH tunnelling using Putty/Plink.
Recently I tried the same to an F7t2 machine, and again today with a rawhide machine that must be pretty close to what is going to be F7t3, Initiating the session with plink.exe and starting gnome-session on the Fedora box tunnelled back to windows, the Gnome session starts up OK, the gnome-panel reacts properly to mouse moves/clicks for a few seconds and then starts running like treacle, only accepting one click every 60 seconds or so, I've checked the same to an FC6 machine and it doesn't occur, I've upgraded to the latest Xming version and it doesn't help.
I *think* the version of Xorg in F7/rawhide now uses XCB underneath Xlib as the transport, I don't claim to know all of Xorg's internal plumbing, but could this be related, or be causing some weird SSH/Xorg interaction?
If I start Xming as a listening X server on the Windows box, then start a command line Putty/ssh connection to the F7 machine, and initiate a Gnome session back to the Windows box, with
export DISPLAY=aaa.bbb.ccc.ddd:0.0 gnome-session &
it seems to run for longer, before meeting the same fate
I realise this problem involves Xming (and windows) as well as Fedora, but it seems to be F7 related, because FC6 and FC3(!) machines still work OK, any other users doing remote X sessions and having problems?
Am Freitag, den 23.03.2007, 17:01 +0000 schrieb Andy Burns:
For several years I've used either Cygwin/X or more recently Xming as an X server for GUI access to Fedora machines from Windows machines, sometimes with XDMCP but mostly with SSH tunnelling using Putty/Plink.
Recently I tried the same to an F7t2 machine, and again today with a rawhide machine that must be pretty close to what is going to be F7t3, Initiating the session with plink.exe and starting gnome-session on the Fedora box tunnelled back to windows, the Gnome session starts up OK, the gnome-panel reacts properly to mouse moves/clicks for a few seconds and then starts running like treacle, only accepting one click every 60 seconds or so, I've checked the same to an FC6 machine and it doesn't occur, I've upgraded to the latest Xming version and it doesn't help.
I *think* the version of Xorg in F7/rawhide now uses XCB underneath Xlib as the transport, I don't claim to know all of Xorg's internal plumbing, but could this be related, or be causing some weird SSH/Xorg interaction?
If I start Xming as a listening X server on the Windows box, then start a command line Putty/ssh connection to the F7 machine, and initiate a Gnome session back to the Windows box, with
export DISPLAY=aaa.bbb.ccc.ddd:0.0 gnome-session &
it seems to run for longer, before meeting the same fate
I realise this problem involves Xming (and windows) as well as Fedora, but it seems to be F7 related, because FC6 and FC3(!) machines still work OK, any other users doing remote X sessions and having problems?
I really think you need to ask this question on the Xming board/list, if there is one.
On 23/03/07, nodata lsof@nodata.co.uk wrote:
I really think you need to ask this question on the Xming board/list, if there is one.
I have already mentioned it to the Xming developer, but as it seems specific to F7 I thought I'd mention it here in case anyone else was doing remote X sessions from F7 to any /other/ X servers and seeing a similar issue.
I do realise that it can get a bit awkward when reporting an issue like which involves non-Fedora software, it would be worse of course if it were a proprietary X server on windows.
On Fri, 2007-03-23 at 17:01 +0000, Andy Burns wrote:
I *think* the version of Xorg in F7/rawhide now uses XCB underneath Xlib as the transport, I don't claim to know all of Xorg's internal plumbing, but could this be related, or be causing some weird SSH/Xorg interaction?
FC7 isn't using XCB yet, still a little sketchy for my taste. I'm aiming for FC8 for it though.
If I start Xming as a listening X server on the Windows box, then start a command line Putty/ssh connection to the F7 machine, and initiate a Gnome session back to the Windows box, with
export DISPLAY=aaa.bbb.ccc.ddd:0.0 gnome-session &
it seems to run for longer, before meeting the same fate
My usual method for debugging this sort of thing is to use wireshark to see which side's network traffic is stopping.
- ajax
On 23/03/07, Adam Jackson ajackson@redhat.com wrote:
FC7 isn't using XCB yet, still a little sketchy for my taste. I'm aiming for FC8 for it though.
Ok, thanks for ruling that out at least
My usual method for debugging this sort of thing is to use wireshark to see which side's network traffic is stopping.
that sounds, errm ... fun, I wouldn't recognize good X protocol over the wire if it slapped me, I'll give it a go though, otherwise I can see that I might end up without remote GUI when F7 comes out, I've put an xming bug into fd.o bugzilla just in case too.
On Fri, 2007-03-23 at 23:02 +0000, Andy Burns wrote:
On 23/03/07, Adam Jackson ajackson@redhat.com wrote:
My usual method for debugging this sort of thing is to use wireshark to see which side's network traffic is stopping.
that sounds, errm ... fun, I wouldn't recognize good X protocol over the wire if it slapped me, I'll give it a go though, otherwise I can see that I might end up without remote GUI when F7 comes out, I've put an xming bug into fd.o bugzilla just in case too.
wireshark has a built-in decoder for X. If you filter on port 6000 then you'll only see the X traffic, and examining packet timestamps should be enough to tell you where the stall is happening.
- ajax
-
Adam Jackson -
Andy Burns -
nodata