[FreeNX-kNX] Windows 98 Compatability

Mon Nov 7 18:27:15 UTC 2005

> Message: 1
> Date: Sun, 06 Nov 2005 18:41:08 -0600
> From: Steve Bergman <steve at rueb.com>
> Subject: Re: [FreeNX-kNX] Windows 98 Compatability
> To: kpfeifle at danka.de, 	User Support for FreeNX Server and kNX Client
> 	<freenx-knx at kde.org>
> Message-ID: <436EA2A4.8000102 at rueb.com>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> 
> Kurt Pfeifle wrote:
> 
> >
> >
> >Which versions of Linuxes? Whcih version of an NX client? Which OS is
> >the NX client running on? Which version of NX servers? *WHICH* 
> >*DIFFERENT* *RESULTS*? What problem with fullscreen resoluton?
> >
> >  
> >
>  From my Friday and Saturday posts, with appologies for reposting so 
> quickly, since there have only been a few posts in between now and then:
> 
> I am running Freenx 0.4.4 on a Fedora Core 2 server.  I have a mix of 
> clients, mostly Linux, but with a couple of Windows (XP) boxes too, running 
> the nomachine client v1.5.
> 
> This is over DSL, so there are occasional connection lapses.  The Linux 
> boxes all just freeze for a short time and then the user can pick up 
> where he left off.  The Windows boxes immediately pop up a window that says:
> 
> "The connection with the remote server was shut down.  Please check the 
> state of your network connection."
> 
> and the user must log back in.
> 
> Any suggestions as to what config parameters I might change to get the 
> Windows boxes to handle connection lapses as gracefully as the Linux boxes?
> 
> I'd wish I could just say "Replace the Windows boxes with Linux boxes", 
> but unfortunately its not an option.
> 
> Thank you for any input.
> 
> -Steve Bergman
> _______________________________________________
> FreeNX-kNX mailing list
> FreeNX-kNX at kde.org
> https://mail.kde.org/mailman/listinfo/freenx-knx
> 
> 
> ============================================================================
> 
> Does anyone at least have any idea why Windows clients would be so much 
> more sensitive to brief network dropouts than Linux clients?  I've been 
> assuming that there must be some difference in the default config, but 
> can't find one.  Is there some other reason?
> 
> Also,  am I on the right list for this question?  I'm unclear as to 
> whether this is an nx list or specifically a kde nx client list.
> 
> Thanks,
> Steve
> _______________________________________________
> FreeNX-kNX mailing list
> FreeNX-kNX at kde.org
> https://mail.kde.org/mailman/listinfo/freenx-
> 
> ============================================================================
> 
> A while back, I found a document explaining the differences between the 
> various connection settings in the nx client.  As I recall, the main 
> down side of setting the connection to "modem" was additional server 
> load.  Unfortunately, I can't seem to locate that document, now.
> 
> On a very lightly loaded machine (7 users) is there really any 
> disadvantage to just setting it to Modem?  I ask because I'd like to 
> have a little more robustness and responsiveness in the event of network 
> congestion, dsl flakiness, etc.  Can anyone point me to an explanation 
> of the various connection settings and what they actually do?
> 
> Thanks,
> Steve
> _______________________________________________
> FreeNX-kNX mailing list
> FreeNX-kNX at kde.org
> https://mail.kde.org/mailman/listinfo/freenx-knx
> 
> 
> ============================================================================
> 
> 
> I have 2 sites using freenx 0.4.4. and nx-1.5.0.
> 
> One is Fedora Core 3.  The other is Fedora Core 2.
> 
> When the client disconnects from the FC2 server by logging out of their 
> Gnome session normally, the window closes and about 30 seconds later a 
> window pops up saying"
> 
> "The connection to the remote server was shut down.  Please check the 
> state of your network connection."
> 
> Also, on the FC2 server, on my 1600x1200 desktop with the client set to 
> "FullScreen", I only get a 1024x768 desktop.  On the FC3 box, I get a 
> 1600x1200 session.
> 
> /etc/nxserver/node.conf is identical between servers.  The software 
> versions are identical.  And the client configurations are identical.
> 
> Any ideas what could be causing the connection messages?
> 
> Thanks for any input.
> 
> -Steve Bergman
> _______________________________________________
> FreeNX-kNX mailing list
> FreeNX-kNX at kde.org
> https://mail.kde.org/mailman/listinfo/freenx-knx
> 
> 
> ============================================================================
> 
> 
> Any insites would still be greatly appreciated.
> 
> Thanks,
> Steve

steve, here are my guesses ... 

network drop sensitivity:

perhaps it's simply a difference in the ip stack between clients. maybe
connections time out more quickly in windows than in linux. i don't know how
you'd go about changing that type of setting, which would likely be system-level
instead of nxclient-level. but it's just a thought. 

can you replicate the problem on the linux clients by disconnecting the network
cable for some period of time? if not, then the ip stack probably isn't to blame ...

i do believe you are on the right list for the questions, because nomachine
periodically monitors the list (or at least used to) and may have insight to
problems with the closed-source nxclient. the server-side libs are open-sourced
so we can peek/tweak, but at least in your case it doesn't look like they're to
blame (because the linux clients behave differently).

though the list name includes "knx" it is also used for freenx, general os
packaging (i try to handle the freebsd ports) and several other client packages
as well.

compression settings:

i don't know the gory details, but i believe the settings remove varying degrees
of round-trips - thereby lowering required bandwidth and resultant latency. none
of my (freebsd) servers are heavily loaded (30-40 users at max) and i certainly
wouldn't set compression to anything less than maximum. i set the compression
myself with web-based configs by using the "nx companion" plugin method.

fc2/fc3 disparity:

as for the resolution i'd assume it has something to do with the differences
between X packages/configurations on the two hosts. i don't run fedora at all so
i'm only guessing.

and for the connection messages ... perhaps some part of the disconnection
(protocol-wise) isn't getting through properly. maybe check your kernel tunables
against each other (with different fedora distributions i'd also assume
different kernels).

ok there, my two cents. i obviously didn't have a single "known correct" answer
for you, which is one reason i never responded to your other posts. another is
i've been away about a whole month and am still playing catch-up.

you might also try #nx on freenode (irc) - you never know, you may get lucky and
find someone awake at the keyboard with more of a clue than myself. well,
finding someone awake may be lucky but finding someone with more of a clue is
more expected.