WideClient problem

[Dick Verburg]

Dear Peter,

We have installed a complete update of PM software as well as FSUIPC, PFC and WideFS software on our PFC simulator. This as preparation for the installation of our new MCP panel. The software we installed are the latest (December) releases.

After this installation we have problems with the WideFS network. The symptoms are that WideClient for one or more of the GC PC’s stops and is no longer active. It happens at various parts of the flight or indeed if you leave the simulator running on runway threshold without any activities.

We are running TCPIP mode, have checked all of the .ini file settings and all of the “trouble shooting” recommendations from your documents and advise from the PM support group. The problem is still there and we just lost the PILOT system. The enclosed file has the PILOT PFD ini and log file as well as the WideClient ini and log file. In addition there is the WideServer ini and log file.

Hope you can give us some advise as to where to look for this.

Thanks,

Dick Verburg / EPST

EPSTfiles.zip

[Dick Verburg]

Dear Peter,

I just had another failure where both the PILOT and the MCP computer lost the WideClient connection. This was after we had changed the ApplicationDelay to 6 (as per another message on this forum). The WideServer log is below:

Does the error 10054 mean anything to you ?

Thanks,

Dick

********* WideServer.DLL Log [version 6.44] *********

Blocksize guide = 4096 (double allowed)

Date (dmy): 05/01/05, Time 15:18:56.515: Server name is MSSERVER

22890 Initialising TCP/IP server

22890 Initialising IPX/SPX server

22890 ServerNode=0.0.3072.62318.44408

23000 Incoming connection Accepted ok (skt=2500)

23000 Incoming connection Accepted ok (skt=2512)

23031 Incoming connection Accepted ok (skt=2524)

23062 Connected to computer "RCDU" (skt=2500)

23062 Incoming connection Accepted ok (skt=2536)

23062 Connected to computer "CDU" (skt=2524)

23093 Connected to computer "COPILOT" (skt=2512)

23125 Connected to computer "ECAIS" (skt=2536)

23437 Incoming connection Accepted ok (skt=2552)

23500 Connected to computer "MCP" (skt=2552)

23672 Incoming connection Accepted ok (skt=2560)

23734 Connected to computer "PILOT" (skt=2560)

4044328 Error 10054: client socket disconnected at Client: removing (skt=2560)

4577422 Incoming connection Accepted ok (skt=2588)

4577453 Connected to computer "PILOT" (skt=2588)

5840859 Error 10054: client socket disconnected at Client: removing (skt=2552)

5916703 Incoming connection Accepted ok (skt=2580)

5916734 Connected to computer "MCP" (skt=2580)

[Pete Dowson]

I just had another failure where both the PILOT and the MCP computer lost the WideClient connection. This was after we had changed the ApplicationDelay to 6 (as per another message on this forum).

Did you try setting the SendScanTime to, say, 50, instead, as also suggested in another message?

The way you say that, you had "one failures" which lost two clients, but you didn't -- one of the clients (PILOT) went off-line for over 9 minutes! The other was half-an-hour later and was the MCP off for less than two minutes.

The WideServer log is below:

Unfortunately, when WideClient has a problem it logs it in the WideClient Log, so what you've shown tells me nothing other than that 6 clients connected okay, then:

4044328 Error 10054: client socket disconnected at Client: removing (skt=2560)

4044 seconds into the session (67 minutes) one disconnected, then:

  4577422 Incoming connection Accepted ok (skt=2588)
  4577453 Connected to computer "PILOT" (skt=2588)

after another 533 seconds (nearly 9 minutes!!!) re-connected. Then:

  5840859 Error 10054: client socket disconnected at Client: removing (skt=2552)

  5916703 Incoming connection Accepted ok (skt=2580)
  5916734 Connected to computer "MCP" (skt=2580)[/quote]

The same with another client, in the 97th minute with it, too, reconnecting after some time -- only 76 seconds this time.

What were the Clients doing in those 9 minutes (Pilot) and 67 seconds (MCP), respectively? Normally, if there's a disconnection, they will reconnect within the second.

Haven't you looked at the Cliient logs at all? It seems rather odd to only show the Server end of things when whatever happened obviously happened on the Clients.

Does the error 10054 mean anything to you ?

It means "client socket disconnected at Client", as quoted more or less direct from Microsoft documentation. In other words, Windows has told WideServer that the Client disconnected.

Regards,

Pete

[Dick Verburg]

Peter,

Yes your observations are (of course) correct. The network guy looking at all this first restarted WideClient on the PILOT and continued the session. After this the MCP failed and he restarted that as well.

We are now setting the SendScanTime parameter as suggested and will let you know the results.

Thanks for the input.

Dick

[Pete Dowson]

Yes your observations are (of course) correct. The network guy looking at all this first restarted WideClient on the PILOT and continued the session. After this the MCP failed and he restarted that as well.

By restarting them without checking the Logs, we can never know what happened. I assume the application programs were restarted too? The fact that one was off for 9 minutes actually implies that something hung rather than any Network problem occurred.

Regards,

Pete

[Dick Verburg]

Peter,

At the moment that a system is hanging (or how does one say this in English), the WideClient application has gone from that system. So it seems like something to do with the network. We set the simulator at a threshold and everything running and check from time to time of all is still running and if it is not, it could be gone for some time.

It runs now for some time after the last changes and we will go for something to eat and come back to see how it is. I will report the results here and all of the settings we changed.

Thanks,

Dick

[Dick Verburg]

Peter,

With the changes below we seem to have fixed our problem where we regularly lost WideClient on one of the PM GC computers.

The server network card is a 3CON Gigabit LOM

The card used for the six PM GC systems (Pilot, Copilot, Ecais, CDU, RCDU and MCP) are INTEL PRO/100 cards

WideClient ini

ApplicationDelay=6 (was 0)

SendScanTime=50 (was 10)

WideServer ini

NoStoppedRestarts=No (was Yes)

Thanks for the useful input and suggestions.

Regards,

Dick

[Pete Dowson]

Peter,

At the moment that a system is hanging (or how does one say this in English), the WideClient application has gone from that system. So it seems like something to do with the network.

Hanging?

There's not been a single case of WideClient actually hanging in six years. How are you determining this? And then how do you remove it? Are all the applications running still responsive, just WideClient "hanging". Quite often an application can "hang" in such a way that it makes WideClient difficult to close -- you have to try to remove the hung applications.

I don't really know what you mean by "gone from that system" -- if this means spontaneously closed, then there will most certainly be a Log file which should be checked to determinate the reason. In fact, even in the unlikely event it is hanging, there will still be a valid Log file which can be examined BEFORE you run it again.

Are you using WinXP or Win98 or what on those Clients?

Please describe more exactly what you mean!! You have me greatly concerned. As I say, what you appear to be describing has not happened at all before. And you have not provided any information whatsoever.

Note that NO FIDDLING WITH ANY PARAMETER CHANGES will make things less or more likely to hang or spontaneously disappear! You are wasting your time and mine hoping for that. Information -- a better description for a start -- is absolutely needed. The Logs ARE important, ignoring them helps not one bit.

:( :( :( :cry: :cry: :( :( :(

Pete

[Dick Verburg]

Pete,

The systems are all WindowsXP home edition. The network cards I desribed before. The problem was that WideClient appeared (I will be carefull now) to have stopped while the application appeared to be still active (not doing anything as there was no connection). Just starting WideClient again apparently made that specific system (PILOT or ECAIS etc.) just function normally again.

I have provided the PFD log and WideClient log and WIdeServer log as attachment to my first message.

If there is anything more you would like me to do, test or provide than you are very welcome to ask and I will do my best to provide it in order to assist tracing what is going on.

Regards,

Dick

[Pete Dowson]

The problem was that WideClient appeared (I will be carefull now) to have stopped while the application appeared to be still active (not doing anything as there was no connection).

"Stopped" as in "Hung"? Not crashed, no error messages, nothing? How did you terminate it, via the Task Manager (Ctrl-Alt-Del)?

I have provided the PFD log and WideClient log and WIdeServer log as attachment to my first message.

Sorry, I cannot find them anywhere. Do you mean in this thread somewhere? The only log extract so far was the WideServer one which I analysed for you.

If there is anything more you would like me to do, test or provide than you are very welcome to ask and I will do my best to provide it in order to assist tracing what is going on.

I'll probably send a revised WideFS build tomorrow (Thursday), with more logging, and maybe some error trapping.

Regards

Pete

[Dick Verburg]

Hi Pete,

The logs are an attachment to the first message in this thread.

The WideClient did not need to be stopped, it was stopped as part of the problem. It was gone from the taskbar etc. Just going to the PFD directory and starting it again did bring that specific system on-line again.

Since we kept a good record of all changes we made, I "assume" that we can go back and duplicate the problem again, so if there is anything you want us to do (when we have time in the simulator, like the weekend or night) we will do it to assist determining the cause of the problem. For the moment it looks a combination of specific network cards and WideFS somehow.

Regards,

Dick

[Pete Dowson]

The logs are an attachment to the first message in this thread.

Ah! Do you know, I've only just found that! :oops: I could have sworn the first one was the one I replied to first. Strange. :?

Unfortunately the logs aren't helpful, except that I note that you are still using TCP/IP. I have switched over completely to IPX/SPX now, on an all WinXP setup. Have you tried IPX/SPX?

I could't tell whether the crashed WideCliet had actually done much before crashing. It would have been more useful to see the WideServer LOG after FS had been closed, as the performance information is at the end.

The WideClient did not need to be stopped, it was stopped as part of the problem. It was gone from the taskbar etc. Just going to the PFD directory and starting it again did bring that specific system on-line again.

So it "crashed to desktop" with no message, nothing? Weird.

I've not had anything like this before with WideClient, so it must be something in the very recent changes. I am putting together another interim test version (6.442, both Server and Client) which will log more performance data, and which also changes some of the defalut parameters again as a result of feedback. I am adding error trapping to WideClient so hopefully I will get a chance to see some details of this crash.

This time, in case I forget anyone, I will be putting up the replacement modules as attachments to an Announcement here. It will be later today, so please look out for it. I'll include details of tests needed and results to be returned.

Thanks,

Pete

[Dick Verburg]

Pete,

I had another ECAIS PC problem this morning to the problem is not totally gone (just lots more time in between occurences).

To clarify either more what actually happens is, you just do not see anyting on the ECAIS (in this example) screen, only that nothing is moving anymore. It than turns out that WideClient is no longer on the taskbar, therefore stopped. No message and nothing shows up on the screen, sympthoms are just that this particular function does not work anymore. When you statup WideClient on that system, all is back to normal.

So there is no crahs, hang (it looks like the system is hanging since it is not doing anything, but as said, just startint WideClient brings it back to business) or other sympthom.

I suggest that I wait for your intermediat release and try that first. Please let me know exactly what I can do to collect the necessary data at the moment this happens again.

Thanks,

Dick

[Pete Dowson]

I suggest that I wait for your intermediat release and try that first. Please let me know exactly what I can do to collect the necessary data at the moment this happens again.

I'm just putting it all together now. Details will be in the announcement and in the ZIP read me. Hopefully, the details I need you to colllect will be in the log. If not, I have a really bad problem.

The thing is, though, because I will be trapping the errors things may look okay. Or they may not. You will have to check the logs on the Clients at intervals, or when anything is supsicious.

Regards,

Pete

[Dick Verburg]

Pete,

Can I just send you the Client logs from time to time or do I have to do something special (close down or something) to get usefull information added to them?

Dick

[Pete Dowson]

Pete,

Can I just send you the Client logs from time to time or do I have to do something special (close down or something) to get usefull information added to them?

Dick

I don't know how you've got your clients set up to respond to Shut Down requests, but if they aren't so configured you can now make the logs include a performance summary by configuring a Hot Key in WideServer to send a shutdown request. Then only things configured to shut down will.

But, honestly, I don't want logs unless there is something BETTER or WORSE, or anything suspicious. Else I'll be overwhelmed and not get anything done! :)

Regards,

Pete

[kzuerner]

Hi Pete,

I have the same experience.

I am running FS4K (flying PDMG 737) on one PC and PM GC and MCP on a second PC. I also have on a third PC running PM GC to display EICAS (or how it is called).

All by a sudden Wideclient exits on my second PC (running GC and MCP). You will not find it in the tasks.

I did not have that problem on my PC running the EICAS.

As I could not see any problem in the Client-Log with 6.44 (have not checked the Server-Log) I took your Version 6.442 and hat the same experience.

Therfore I send you attached the two logs. Maybe you can trace the problem now.

Regards

[kzuerner]

I seem to have some problems to attach the log.

I try it again.

Regards

Logs.zip

[Pete Dowson]

All by a sudden Wideclient exits on my second PC (running GC and MCP). You will not find it in the tasks.

Aha, that makes 3!

Therfore I send you attached the two logs. Maybe you can trace the problem now.

I hope so. They've not arrived yet, and I have some chores to do, but I will check them later tonight. Thank you!

[LATER]

AhI see you attached them instead of emailing. Okay, they were short enough. Thanks.

Pete