Strange FSUIPC Multuiple Connections suddenly

[hladamjr]

I have been using the registered FSUIPC 3.5 for several months now. I have SB3, Widefs Server, ASV, and Servinfo running running on one PC and FS9, Servinfo and Widefs client on the other. Recently I will be flying online and be enroute with everything functioning fine when suddenly FS will hang or freeze. What I have noticed is that normally I show 1 client connected via Widefs when FS9 hangs or freezes I now have noticed that the number of Widefs connections will increase to as many as 11 connections. I use a hardware router/firewall for WAN access and also Zonealarm Pro on both PC's. I have the approiate ports opened and everything works fine most of the time with the exception of when I get multiple connections through Widefs.

Just wondering if this has been seen before and if it is a "known problem" that has been encountered and I missed it in the Forum.

If I shut down everything and restart things will be fine and I can start a new flight and usually be fine for the 2-3 hour flight. It happens usually once every 5 flights or days nothing really regular as I can see.

I am very experienced in network architecture and configurations and see nothing out of the ordinary in log files for the firewalls, so I am somewhat stumped.

Any help , insight or related expierences are appreciated in advance.

Henry Adam

[Pete Dowson]

Recently I will be flying online and be enroute with everything functioning fine when suddenly FS will hang or freeze. What I have noticed is that normally I show 1 client connected via Widefs when FS9 hangs or freezes I now have noticed that the number of Widefs connections will increase to as many as 11 connections.

This phenomenon is clearly described in the WideFS documentation. The freezing (or slow running) of FS is likely to be the cause, not an effect. What happens is that WideServer isn't able to respond to your client(s) often enough, if at all, and they timeout and attempt to re-connect. The count increases because the Client's timeout is shorter than the Server's -- eventually the server will time each inactive connection out and the count will decrease, but for that to happen the FS freeze or whatever it is has to clear.

Just wondering if this has been seen before and if it is a "known problem" that has been encountered and I missed it in the Forum.

You missed it in the documentation for WideFS.

If I shut down everything and restart things will be fine and I can start a new flight and usually be fine for the 2-3 hour flight. It happens usually once every 5 flights or days nothing really regular as I can see.

One possibility, if it occurs after a number of hours of running FS, is that you have a memory leak. Once the real memory gets used up through such problems, things get continually swapped out to disk. When this happens it also happens to the Winsock buffers used for the Network, and that clogs up too. It will gradually get worse until FS freezing is possible, if not an outright crash.

The most common cause of memory leaks is related to LandClass files in sceneries. I had such problems for a while and had to move a load of landclass BGLs to their own Scenery folder with no Texture folder. I think there's a program which will help you do that automatically -- FlightSim Manager or similar.

Also, if you are running the original FS2004 still, update now to FS9.1. That fixes a load of other possible memory leak problems (but not all, unfortunately).

Regards,

Pete

[hladamjr]

PEter

Thanks for the speedy reply ! I'll review the Widefs docs. and look at the timeout settings first. If that resolves it then I don't need to move and BGL files. I'll post a reply with what I find and what resolved the issue.

I'll change timeouts if need be first and fly with that before changing anything else as I never make multiple changes at one time. Helps to know what resolved things or was the issue. Also it make it simple if new issues occur after making a change.

It may be a few days before I can post the results as I want to be sure of the resolution. I will post and interum findings or issues if the occur.

Thanks again for the prompt reply.

Henry

[Pete Dowson]

I'll review the Widefs docs. and look at the timeout settings first.

The timeout settings should really be left to default. The current values have endured many years will good service. If you start dabbling in some of those numbers you'll find it a never-ending job trying to balance everything.

The timeouts are not the cause of anything. They are merely a safety valve, to try to recover from lost connections.

I hope you haven't changed anything unnecessarily in the WideFS INI file config settings? Generally only the protocol and (optionally, now, with WinXP systems) the ServerName is needed.

Regards

Pete

[hladamjr]

Peter

I downloaded and registered the FS ADdon Manager you mentioned. I am slowly weeding through things with it. I have removed many addon A/C that I don't use often. I haven't gotten to the scenery or AI part yet ( a little more reading required) I make another flight today aftewards and again FS hung and I have included the WideFS log from that session as an attachment. I noticed alot of sockets that get disconnected after a period of time.

Any insight or direction is again appreciated.

Henry Adam

WideServer.zip

[Pete Dowson]

I downloaded and registered the FS ADdon Manager you mentioned.

I don't know the Addon Manager -- the one I have is definitely called FlightSim Manager. See http://www.ranainside.com.

I am slowly weeding through things with it. I have removed many addon A/C that I don't use often.

Whilst unused aircraft add to the clutter on disk and may slow down the opening of the aircraft selection menu, none of those will be contributing to the lack of stability in your installation.

I make another flight today aftewards and again FS hung and I have included the WideFS log from that session as an attachment. I noticed alot of sockets that get disconnected after a period of time.

That log shows a sorry state I must say. Much more worrying than the disconnections are the Sumcheck/Length errors. This is actual corruption of data coming across the network! That is really really bad and should never be seen! These are the lines you are not looking at:

195437 **** ERROR! Sumcheck or length fails on received socket 14840 block, len=370 (time=0)

They are occurring sporadically through the log segment you sent. The **** ERROR! part is meant to draw your attention to them, so I'm surprised you picked out the disconnections instead, which are never causes but results.

The disconnections logged as:

10397984 Error 10053: client socket disconnected at Client: removing (skt=22776)

may have rather more relevant information in the Client Log, as it is the client that did the disconnection, as it says.

The fact that these errors are spread sparsely over the 4 hours 36 minutes shown in the log, and not simply bunched up after several hours, may be more indicative of something else wrong other than the memory leaks which I hypothesised before. Unfortunately the performance reports from WideServer are missing, presumably because FS hung and not closed? Did you end the FS9.exe process via the Windows task manager (Ctrl_Alt_Del)? If so, and WideServer still did not summarise and close the log, this means that something serious in FS's main message processing thread was hung.

Perhaps, if you closed the WideClient end the performance log from there would show something.

With stuff like data corruption in the blocks being received by WideFS I would normally suspect the Network -- hardware or driver. But this alone would never hang FS. Combined with FS hanging I would instead suspect the PC itself -- interrupt conflicts or the memory chips failing intermittently. If the system always used to perform well and has only started playing up recently, then I would certainly suspect something on the motherboard, with memory being the most likely. If you have more than one memory stick try without one at a time to determine if it is memory and if so which stick.

If it has always been a problem on this system, then also check the interrupt assignments (IRQs). In particular check that the video card is not sharing the same IRQ with the Ethernet adapter.

Anyway, as I said, I don't think the FS hanging is at all related to the WideFS problems. It looks like they are both results of a common cause, however.

Regards,

Pete

[hladamjr]

Pete

Looks like I have some digging to do this weekend. I have noticed a slight performance degradation lately. I will look at memory as a possible cause. I have 2 GB of RAM in the computer that runs FS so lack of memory shouldn't be an issue. I will do some testing.

The performance log is missing due to the fact that FS just shutdown after hanging and logs did not get updated/closed properly I assume. I will also look at client logs also.

If you can yell me which logs are relevant and I can include them going forward and also what other information may be helpful that I should include.

Thanks again for you diligence and always helpful assistance.

Henry

[Pete Dowson]

If you can yell me which logs are relevant and I can include them going forward and also what other information may be helpful that I should include.

Well, for WideFS there's only the logs from Server and Client -- unless you've been changing anything in the INI files, but in the latter case I'd ask you to revert to default values first and re-test.

The underlying problem causing FS to crash is not something I can diagnose I'm afraid. When I've had such problems it is most a matter of trial and error. If your system has been good for ages then suddenly starts misbehaving, then, assuming no rogue programs (worms, viruses) have got in (check that, of course), then some sort of hardware fault or deterioration is the major suspect.

Software doesn't deteriorate or develop new faults with no changes being made to it. There is of course the possibility of software corruption (files not being right) though that would also normally only arise from a hardware glitch -- though maybe something you've installed is responsible? For FS the usual things causing such hangs are corrupt texture files, corrupt gauges, corrupt weather files. But these are not likely to be the cause of the errors seen on the Network. As I said, the only common thing I can see would be memory and interrupts or, worse, something intermittent on the motherboard like a hairline crack in a track.

With as much as 2Gb of main memory the system may perform perfectly for many things. It takes something which really does use a lot of memory, such as FS, to hit it all and hit it enough. Try removing each memory stick in turn (I assume the 2Gb isn't all on one stick? ). Give each variation a good run, only passing the removed stick if you still get the hang.

One other thing to check is the temperature of your processor and video card. At this time of the year (in the Northern hemisphere) with most folks turning their heating on and cutting ventilation (draughts) out more it is surprising how much hotter PCs can get. However, in my experience most such problems tend to result in either a solid PC hang or a spontaneous hardware reset and re-boot.

Regards,

Pete

[hladamjr]

Pete

:D Good news ! I am back up and running I think with no errors. Now the bad news ..... :cry:

I began digging and noticed more performance problems. I ran a chkdsk and found numerous HD problems. I ended up getting a new drive and reloading everything, XP, FS etc, etc. I have things almost complete and can connect fine with no errors.

So I think I was getting data corruption at the HD leval basically.

In any case thanks for all the help and keep that one in mind as an outside possibility :idea:

Henry

[Pete Dowson]

:D Good news ! I am back up and running I think with no errors.

... I think I was getting data corruption at the HD leval basically.

Ouch!

Glad you sorted it. Thanks for letting me know!

Regards,

Pete