Hi, I had 2 out of synch games yesterday.

Unfortunately I am not 100% sure which game data version it was. I suppose the first one was 3.7.3 and the second one 3.7.0 since I reinstalled the game data and if I remember correctly u had stopped the release of 3.7.3 at that time already.

(Probably) 3.7.3:
Civs: Huns, Spain, Bengalis, Khmer, Frank, Goth, Vikings, Malians

(Probably) 3.7.0:
Civs: Chinese, Tatars, Hindustians, Vikings, Romans, Japanese, Celts, Lithunians

Before the desynch Romans have attacked with Elite Centurions, which other players already have reported to cause desynch issues. It might have to do with the trample damage??

In both cases resuming was impossible since always one player couldn't join.

Edit: I think u can ignore the version numbers. I probably got confused with the version numbers of the patch and game data since they are similar.

One more OOS today (again no Romans involved):

Player OOS was NerveImbecile I think.
Civs: BENG, SARA, GURJ, BOHE, DRAV, BRIT, HUNS, PORT

I have 2 OOS files (if they are important anyways), since we tried resuming 1 time but this time I was suddenly disconnected, so I might have been OOS myself. Anyways, I include both OOS files. I only have 1 rec file though, the second shows reading error.


Oh BTW (off-topic): When watching this rec (haven't tried it with others yet) my reso switches to 1024x768. Even when I change reso and restart the rec it switches back. Didn't have that issue before, may it have to with the latest update? Edit: It seems to be a permanent issue now. Also when watching other recorded games, reso will switch to 1024x768.

Offtopic 2:
Is there any way to fix the issue which happens a lot lately that u can't restore a game because always 1 or 2 people can't enter the room? Is there any way we can assist you finding out what the problem is? I don't think it's shown in the Event Viewer nor in the Voobly.log, right?

Do you have match links for us where this happens? e. g. that players cannot join or were getting "unable to join".

0x001efe96 crash it is due to 1.5 rc patch.

I think it mostly happens after a desynch error, don't remember right now if it also happened after someone just dropped. In my first post in this thread there are 2 recs where this happened. Apart from that I don't remember the games, but I can post more if it happens again. I also don't know what happens with the players who can't join, if they receive an error message or anything. I just know that in the pregame it seems like everyone joined, but after a few seconds one or two players will turn red again and it seems like they are downloading. Then they will appear green again for 2 seconds and so on.

Ok just had it again. Classical Romans Desynch error and after it Twister couldnt join the restore game.
Dont know if he saw an error message, but from game room it seemed like he joined and instantly got kicked out (and again and again after several tries).

1.6 out of sync

player out of sync according to game: 12_ghoolam (Koreans)

civs: Mayans, Sarascens, Bulgarians, Byzantines, Incas, Burmese, Slavs, Koreans

No specific player mentioned "an error occurred"

Civs: Romans, Turks, Persians, Malians, Byzantines, Hindustani, Bengalis, Britons

Is someone looking at this? They are very regularly desyncing again and people start to play 1.5 more...

No specific player "an error occurred"

Civs: Persians, Bengalis, Berbers, Bulgarians, Ethiopians, Sicilians, Mongols, Cumans

Yes, as far as I know, Mikael is checking stuff but it's a very hard process since reproducing is almost impossible.

Did a bit of web research on the issue today and apparently this OOS error is indeed a nasty problem with multiple possible causes and not only occuring on Voobly, but also on Steam and in Mac/Linux environments. However, I found this interesting report on a specific type of OOS issue the Forgotten Empires devs fought with (with technical details about the cause and the fix):

"Age of Empires II HD: Patch 4.6 – Out of Sync Hotfix

Dec 10, 2015

Many of you who have been playing Age of Empires II HD lately, may have noticed an unusual amount of “Out of Syncs” going on. These errors happen when the game “world” looks different for various players, making the game no longer synchronized. This bug was introduced with the launch of The African Kingdoms, but we can happily announce that the bug is fixed and is now a relic of the past.

Your game will be automatically updated to the latest 4.6 patch. If you wish to force an update, just restart your Steam client.

aoe2_HD_outofsync_patch_46_released
Technical bits and bops

So, what exactly was going on? After a lot of digging by the team (especially Morness, lead programmer at Skybox Labs), we found out that flares could sometimes cause a memory overflow. At this point, a cascade of errors is unleashed.

Each flare is assigned an object ID, however, since we added player-colored flares, something went wrong and the flare objects could overwrite IDs 1-8.
These object IDs (1- were almost always trees
At this point, if a team would flare, the flare would only exist for their team, and not for the other team. And thus, for the whole team, a certain tree would be corrupted (still visible, just not usable. However, the other team would still have the normal tree in their world.
Once this tree is about to be cut, one team sees an invalid tree and the other team sees the valid tree, at which point the game state is no longer the same for both teams and the game goes out of sync.

How come this bug was in the release?!

It’s a crazy bug for sure and actually one we did not encounter during our testing. If you’ve made it through the technical explanation above, the reason is actually pretty simple. Most of our QA team are either in the same room or use Skype when playing together. In both cases, flares are rarely used.
Follow up

Due to the harsh nature of this bug, we vastly upgraded our logging mechanism for Out of Syncs errors. Since our logging started, last week, 7.6% of the games went out of sync. We’ll keep an eye on that number but with 3 OOS causes patched in the last 7 days, we’re sure to see that number drop to an all-time low."


2 questions:
1) Is it possible that flaring might also be a main trigger for the desynch issues in v1.6?
2) If that's not the reason, would it be possible to introduce a logging mechanism for OOS games like they did and do you think it might help identify the (main) causes of the error better?

0x001efe96 crash it is due to 1.5 rc patch.

Those 3 games I reported above were definitely not 1.5 RC...

All 1.6 RC + 1.6 Game Data

I agree with Sickle. I never played 1.6 + 1.5 Patch before, nor do I ever see games with this setting. Either 1.6/1.6 or 1.5/1.5. The outsync almost exclusively happens with 1.6/1.6.

0x001efe96 crash it is due to 1.5 rc patch.