Fulda scenario hang

[griffin115]

I had just downloaded William's new maps and loaded up the Fulda Frontier of Freedom scenario to look at the map. So I thought, while I'm here, I'll set up the units and play a few rounds of orders to see how usable the new map was. All went well until exactly on the four hour mark which is a point when reinforcements are expected. The screen suddenly stopped updating and would not respond to the keyboard or mouse.

I loaded task manager and checked the process. It was running flat out on one processor and the memory usage was incrementing rapidly. It kept updating up to 1,583,520Kb and then stopped updating and dropped the CPU load to zero.

When I tried to kill the process, Windows said that it was waiting for user input. The keyboard and mouse were still unusable so I had to kill the process.

I'm running Windows 7 64 bit on an i5-2500K with 8Gb of memory and a Nvidia M570 graphics card.

Peter

[cbelva]

Hey Peter, Can you attached the scenario file and the saved game file taken just before the system stopped responding. This sounds like an old bug that was fixed some time ago where pop up messages were being hidden. I would like to see if I can trigger it. It I can, Rob should be able to fix it (it may already be fixed because there has been a lot of changes to the game since 2.03a). Thanks

Never mind about the scenario in question. I see which one it is. However, the saved game file would be nice to have.

[CapnDarwin]

As a sanity check, what game version are you running?

[griffin115]

I am running 2.03a. I don't have any save games - I was just going to play a few turns to see how the replacement map felt and then a few more turns ..... I'll try to reproduce it and save the game as I go.

[cbelva]

I played the scenario out to hour five yesterday in 2.04 beta and could not reproduce. If you could trigger again with a saved game it would be helpful.

[griffin115]

I have been doing some more testing and the outcome is that there is no evidence - the crash is not reproducible by just providing a save game. The crash has occurred each time but only after several hours of in-game time. The crashes are as follows:
1. The original crash - screen became unresponsive at the three hour point as detailed above.
2. I reran the scenario but took a save just before the 3 hour point. The game continued to about 3hrs 20 mins and then hung with a black screen. When I killed the process and checked the save game, it had a zero file size.
3. I reran the scenario taking save games and checking their size. I went to bed at the 3 hour point. When I loaded the save game the next day, it continued for several hours and then hung again. I reloaded from the most recent save game and the scenario continued to the finish.

The fault appears to arise only after several hours of in game time. I'll check it out again when 2.04 arrives.

I haven't played any of the large scenarios before and learned a lesson with this one. I thought that emphasising counter battery fire early on might save me some grief later. Never again. When I inspected the results at the end of the game, only one artillery unit was killed. I could have spent that time killing tanks!

Thanks for the response.

[CapnDarwin]

We are aware of CB fire issues and have them on the list of things to correct with the arty overhaul with 2.1. We know it was working at one point, but something is off the rails.

[cbelva]

I have been running test to see if I could duplicate Peter's problem. The other day I ran it twice playing as the Americans using the stock map that came with the game out to just past 5 hours and it never hung. I was also playing under 2.04 beta. I today ran the scenario allowing the computer to play both sides and using Williams map. It hung up on me when playing with 2.03a build after 7 hours 23 minutes. I ran it a second time under 2.04 beta and it hung on me at 4 hours 11 minutes of game time. This time when I clicked on the map, I had a Secure Transmission message try and open. The narrative windows were whited out (see pix). It appears that this pop up message may be the culprit. I have sent a message to Rob about this. I don't know if the problem is the map or the scenario. It was not one of the official game scenarios. It was created by one of the beta testers before release and submitted to be included. I am unsure if this scenario has received the benefits of the updates.

[griffin115]

I tried it again a final time with the original map, although I didn't expect any difference. It hung at about 4hrs 20 mins. I reloaded the autosave and the game continued for several in game hours before I abandoned it. It's easily resolved by using the autosave. Now I have to maintain the discipline not to use it. [:)]

[CapnDarwin]

Can both of you tell us what your cpu, or, and ram memory are? It sounds like we are hitting a memory cap limit possibly with map markers.

[griffin115]

I'm running Windows 7 64 bit on an i5-2500K with 8Gb of memory and a Nvidia M570 graphics card.

[CapnDarwin]

Thanks. We'll put some eyes on it and see if we can figure out the problem. Has any other large scenario crashed on you?

[griffin115]

I have been playing the smaller scenarios, I haven't played any of the other large ones.

[CapnDarwin]

Okay, thanks. Let us know if any other scenario breaks for you.

[cbelva]

I think that this bug is specific to the Fulda scenario. This is not one of the stock scenarios. It is user made even tho it was released with the game. I can get the hang bug to work every time I play this scenario. It triggers between 3 hours and 7 hours of game time. I have ran numerous other scenario and cannot get the bug to trigger.

[IronMikeGolf]

I am running this scenario as computer-vs-computer. First run, it crashed sometime after 6 hrs 30 min. Event log said GDI32.dll threw the exception. Trying agains and will save it if it gets to 6 hrs. I have enabled User mode crash dumps for this run.

[CapnDarwin]

Jeff, thanks for the test and the error message. The GDI error supports my theory of a map marker generated crash. The large number of units coupled with the large number of map markers may hit a tipping point in memory allocation or addressing. Hopefully Rob can zero in on this issue.

[cbelva]

Hey Jeff,
Rob is back and has been looking into this. I believe he was able to trigger this bug himself the other night. We will have to wait on the verdict from him as to whether he can fix it. I think that the bug is related directly to this scenario. There may be data corrupted (I don't know for sure). However, this is the only scenario that we are seeing this bug right now. It is not one of the stock scenario created by MR for the game. It is a user made scenario (made by a beta tester) that was included in the release. We will let you know what the final determination is. Thanks for your post.

[IronMikeGolf]

Failed again. Sometime after 5:44 into it. I did not set autosave and when you do computer running both sides, the menus are disabled when the game is paused. Except for Help.

I attached a zip of the dump file. The error in that is 0xc0000005, "The thread tried to read from or write to a virtual address for which it does not have the appropriate access".