wine users - Mar 2002 - How to troubleshoot network woes with Wine?

The app which I am currently trying to get to work properly (Cinema4D
network rendering client - it does work 90%, depending on how you look
at it) does some weird things, like receiving jobs from a server,
computing them, and sending the rendered scenes back to the server
(actually I don't know if it's sending them back, or if the server is
fetching them; I don't even know what the general mechanism is: ftp,
http, smb . . . ). Now, except for the occasional crash (sometimes after
a few hours of rendering), everything works except the sending back of
rendered scenes to the server. Exasperatingly, it sometimes *does* work,
i.e. some scenes do get sent back sometimes . . . 

I don't have any idea at the moment how I would go about narrowing down
the possible reasons for this. How do you tackle networking stuff with
wine in general?

Regards, Frank

(it would be frustrating if I had to pack in so close to the finishing
line . . . )

On Fri, 8 Mar 2002, Frank Joerdens wrote:
> I don't have any idea at the moment how I would go about narrowing down
> the possible reasons for this. How do you tackle networking stuff with
> wine in general?
>
> Regards, Frank
>
> (it would be frustrating if I had to pack in so close to the finishing
> line . . . )
Well, in general, all Wine services are actually provided by the
underlying *NIX OS, so if your windows app will be able to connect to
something, a native Linux app would too.  Windows ftp, telnet, and http
use winsocks, which in turn uses *NIX sockets, so AFAICT these clients
behave very much like the native *NIX ones...

All I know to do to try to get my teeth into an unknown app is to run a
relay trace and try to see what Windows services it is trying to use
when it fails.  Well, usually I use $relay to filter out some of the
more repetitive calls.

[whit@giftie whit]$ printenv |grep relay
relay=-relay=rtlentercriticalsection:RTLleavecriticalsection:rtldeletecriticalse
ction:initializecriticalsection:interlockedincrement:interlockeddecrement
[whit@giftie whit]$ tty
/dev/vc/5
[whit@giftie whit]$ cat /dev/vcc/5|fold|trail >oof

I haven't checked that to make sure they are all there and spelled
correctly, but maybe it will give you an idea.

Maybe somebody else knows the app and has a better idea.

Lawson

Q:  How many IBM CPU's does it take to execute a job?
A:  Four; three to hold it down, and one to rip its head off.

I have a *probable* culprit: This app uses ws2_32, aka Winsock 2, which
is, according to a thread from November in wine-devel, only partially
supported. The fact that I have partial network functionality (i.e. it
even works OK *sometimes*) might be considered kind of consistent with
the fact that Winsock 2 is only partially supported. The only real
warning I get with

wine --debugmsg +ipaddress,+winsock,+dll c:\\Program\
Files\\C4DNET_R73\
> PC\\C4D_NET_Client.exe
is

warn:winsock:wsaErrno errno 22, (Invalid argument).

Does anyone have any ideas as to this error? Are there any current
Winsock 2 patches floating about I could play with?

Regards, Frank

On Fri, Mar 08, 2002 at 07:43:41PM +0100, Frank Joerdens
wrote:
> I have a *probable* culprit: This app uses ws2_32, aka Winsock 2, which
> is, according to a thread from November in wine-devel, only partially
> supported. The fact that I have partial network functionality (i.e. it
> even works OK *sometimes*) might be considered kind of consistent with
> the fact that Winsock 2 is only partially supported. The only real
> warning I get with
> 
> wine --debugmsg +ipaddress,+winsock,+dll c:\\Program\ Files\\C4DNET_R73\
> > PC\\C4D_NET_Client.exe
> 
> is
> 
> warn:winsock:wsaErrno errno 22, (Invalid argument).
> 
> Does anyone have any ideas as to this error? Are there any current
> Winsock 2 patches floating about I could play with?
One other thing:

frank@limedes:~ > cat /proc/net/sockstat
sockets: used 68
TCP: inuse 29 orphan 0 tw 54 alloc 29 mem 3
UDP: inuse 11
RAW: inuse 0
FRAG: inuse 0 memory 0

The app opens sockets galore (68 after less than an hour of uptime) on
the assigned port, which is probably also some kind of misbehaviour that
is part of the problem . . .

Regards, Frank