Dr Watson error on load

Not sure is this ties in with the occasionally seen rendering issues, but on my machine when I've maybe had Fantasy Grounds open and closed a few times, maybe using two sessions (local player connecting to local server) now and then, eventually I start getting an error message when loading the second instance, or even when loading the first instance once.

The message reads "The procedure entry point SymSetSearchPathW could not be located in the dynamic link library dbghelp.dll"

Attachment 1876

Searching the forum posts, I couldn't find any mention of this. As I see it quite regularly, I therefore assume it's probably due to my multiple restarts and usage of FGII, or the apps I also run as an FGII developer. Photoshop might interfere (big memory load), although even after shutting it down the problem persists.

The other thing is I'm running with an NVIDIA GTX 460 SE card, which might have it's own set of issues to deal with. Indeed, when I pulled up Speccy to post some machine stats, I managed to get an odd see-through background issue, which might be the best clue I have for now.

Attachment 1877

In any case, a full reboot sometimes fixes the problem, and I can normally start one instance of FGII, but I never used to have problems bringing up two or three instances of it before. Just wanted to post here to see if anyone else ever sees this issue ?

That function is not used in the FG primary code base, so not sure where it's coming from. From the MS developer's web site, that function appears to be used for looking up MS common library debugging symbols.

The see through issues I usually only see with driver issues or low graphics memory situations on my machines.

Regards,
JPG

The graphics display issue is a long standing bother with my new card - might not be related, who knows.

The Dr Watson error only seems to come with second / third windows opened... but they were always player windows (as the first window is the GM window, of course). So on a moment of inspired genius, I deleted the campaign cache and reloaded. Viola, it works now.

Will keep an eye for when it happens again and see if this is was the root cause or just a symptom...

Wow - just our current session, one player had some dice rolling lag, and then suddenly my server session crashed, with a C++ runtime exception (didn't manage to snap it), followed by the usual failed to load exception (which I think is a DirectX failure, at a guess).

Probably still a "my graphics card" thing, but it left me with a corrupt DB file (lots of character sheet style tags in the root of it ??), so had to pull up the backup file and lose the last 10-15 mins of the combat. Quite annoying :(

Any information on what actions were attempted/occurring at the point of the crash would help point me to a section of code to review.

As I work on v2.8, I have been continuing to tighten up many aspects of the code to reduce any crash issues. If it is a video/driver issue, I won't be able to do much other than try to help debug.

Thanks,
JPG

We were in the middle of combat, and waiting for my PC to roll her dice. She was doing this, but they were taking a long time to show up at the other player's screens and my screen. Finally she may of logged out, and at this point, my server crashed.

All that comes to my mind around this is the dice seem to be the thing that goes a bit off whenever I'm having graphical issues. E.g. they render funny in all the screens when I'm about to have a Dr Watson error. I suspect this is because they lean on a DirectX library, which seems to be the thing that starts failing (until I reboot) when I get these issues with my graphics card. It certainly stops me loading any games afterwards, as OpenGL / DirectX are all busted.

The only reason I haven't said "let's just forget it" from an FGII perspective is I've not seen these errors occur during a session where FGII hasn't been run at some stage. That said, I normally run FGII about 3-4 times a day as I'm actively developing it, so short of stopping FGII development for a few days...

I'm a total novice on Windows / DirectX / DLLs / crashes, so not really able to offer much more technical detail, but happy to try and provide some with instruction on how to get it !

phantomwhale - A thought.

If you are actively restarting FGII several times a day as your developing before running a live game, be aware that continuously reloading the ruleset using /reload from within FGII will eventually lead to a C++ Runtime error upon restart or while FGII is running.

It used to be much worse however JPG's optimisations have since reduced their frequency. However as of 2.7.4, they haven't been 100% eliminated.

Its therefore worthwhile shutting down FGII and closing the application every 3-4 reloads to ensure a clean restart. I would especially do this, prior to running a game where multiple clients will be connecting.

You changed your avatar. I fear change...

Thanks for the input - yeah, I more often than not restart rather then reload (a lot of the testing is client's connecting to server's now, so /reload all has limited benefits). And I always do a fresh reboot of both computer and router before any actual game session !

I wonder if FG is using some portion of the DirectX library that does not come up in other applications, and thus raises the issue. Based on what was happening, I would usually looking at either graphics code or the network code handling dice.

Unfortunately, the DirectX portion of the code is very complex, due to the unique physics simulation that the original developers created. So, my experience with that part of the code is limited.

For the network code, it wouldn't even get to the point where dice are rolling if there was something wrong in the networking code.

I'm a bit stumped on this one.

Do you happen to have another machine lying around to double-check the behavior?

Thanks,
JPG

An old friend I met up with had a spare NVIDIA 8800 GT which I am now using whilst I send off my GTX 460 SE for investigation.

Just loaded up four FGII instances, one server and three connected clients. No problems whatsoever.

Looks like my GTX 460 (Gigabyte) graphics card was having DirectX issues (and OpenGL issues, I think) - so will be sending it in for testing. FGII is therefore most probably innocent (as suspected) !

Spoke too soon - replacement graphics card is also exhibiting similar problems to the old card. Which means... the problem isn't a graphics card problem at all ?

Have tried other graphics drivers (including very old ones for the 8800 GT) but no avail. So the current thought is - bad OS install. Something in windows is screwy. So I've gone with a brand new build.

Results so far are... semi-encouraging. Nothing seems too flaky, and some of the core problems are gone. However I am still getting errors from FGII when opening several windows and wanted to check if THESE errors are more "expected" ?

Brand new install of FGII, loaded up a new 4E campaign, and connected two clients (locally) to it. Not a problem :

https://www.fantasygrounds.com/forum...1&d=1302583249

Then I tried opening a fourth client and connecting to the same campaign. Here, we see a problem :

https://www.fantasygrounds.com/forum...1&d=1302583249

Zooming in on the error message :

https://www.fantasygrounds.com/forum...1&d=1302583249

I've seen this error before - in fact I seem to recall pulling out that DLL from a Microsoft engineering distribtuion (as it's a debugging DLL) and adding it to the FGII path previously, which might simply start giving me the OTHER message from the start of the post ?!

So my concern is I'm still seeing SOME problems, releated to multiple windows open, but it's still too early to tell if it's the SAME problem exactly, as with my old OS install I'd sometimes get errors after opening two, or sometimes even just one window !

Has anyone seen this before ? Or do other people get errors trying to open four or more instances (not a common usage of FGII, granted) ?

Regards,
Ben (-PW-)

It looks like you are running more than one client on the same machine. While FG doesn't prevent you from starting the application, it is generally a bad idea. The problem is that there are shared resources (i.e. cache) being written to and read by all the clients at the same time, which can cause a lot of strange and unpredictable issues.

You will not see the same problem with host and client on the same machine, since they write to different directories.

Regards,
JPG

I most certainly am running multiple clients on one machine (and I DO see cache write errors too !)

Mainly I've been doing this for local testing, and documentation (how to mimic six players logged in without having six machines or actual players handy...), but it's normally with just two sessions.

But like I say, was having trouble even holding two sessions open without the Dr Watson error popping up before. Bit worried this install of Windows will go the same way, and is only managing three now as it's "brand new and fresh". And I normally only test with two (client and server) as this is fine for cache sharing, and lets me mimic most behaviours required.

-PW-

Other than multi-client testing, I fudge showing multiple players logged in on my example graphics by having one client open multiple characters.

Cheers,
JPG

That... would be much easier :)

Quote:

---

Originally Posted by moon_wizard

Other than multi-client testing, I fudge showing multiple players logged in on my example graphics by having one client open multiple characters.

Cheers,
JPG

---

However, the eagle eyed will notice one name in bold whilst the rest aren't! ;)