[EXTREME PRIORITY] Crash the game when you start a new game or try to start saving

[DedKamazy]

Version of game 0.9.4.0
OS Linux Mint 18.2 Cinnamon 64-bit
Kernel 4.10.35-generic
CPU Intel Core i5-3450 3.10Mhz
RAM 16Gb
Nvidia GM206 [GeForce GTX 950]

Simple Description Crash the game when creating a new game or loading an old one.

Detailed Description When you create a game, the game is closed in any place when the world or objects are generated. Seldom can stand in any place of creation. The HDD stops being active and all 4 cores are 100% loaded.

Reproducibility Constantly emergence, it is impossible to start the game from 20 attempts of any successful.

Player.log.zipSave.zip


P.S. If you need any additional information, please specify and I will certainly provide it! Thank you in advance!

[lucy]

this is probably the same problem https://forum.planet-nomads.com/thre...-t-launch-game
but they didn't even notice it yet :/

[Menzagitat]

...
but they didn't even notice it yet :/

I am sure they noticed it. They notice most spam posts.
@devs: this must be deleted too: https://forum.planet-nomads.com/thre...ll=1#post18613

P.S. If you need any additional information, please specify and I will certainly provide it! Thank you in advance!

@devs: A more polite request you cannot get.

I had a friend who asked me if he should buy the game.
At that time the game had many bugs and no game play and I was unsure what to answer to him.

Then he asked me how is the customer support.

[Goose]

Version of game 0.9.4.0
OS Linux Mint 18.2 Cinnamon 64-bit
Kernel 4.10.35-generic
CPU Intel Core i5-3450 3.10Mhz
RAM 16Gb
Nvidia GM206 [GeForce GTX 950]

Simple Description Crash the game when creating a new game or loading an old one.

Detailed Description When you create a game, the game is closed in any place when the world or objects are generated. Seldom can stand in any place of creation. The HDD stops being active and all 4 cores are 100% loaded.

Reproducibility Constantly emergence, it is impossible to start the game from 20 attempts of any successful.

Player.log.zipSave.zip


P.S. If you need any additional information, please specify and I will certainly provide it! Thank you in advance!

Hi DedKamazy,

Thank You for report and sorry for a late response. I've resent it to right person and hopefully it will be fixed soon. I will contact you if more info is needed.

Thanks !

[liamdawe]

FYI, I've been posting about this issue since August: https://steamcommunity.com/app/50405...9519988560040/

Please see my latest post there. The working database file for saved games becomes malformed.

[Ehvis]

Just gave it a try as well. Same for me. Player.log gives me a whole bunch of "SQL logic error during BulkExecute" errors.

The only report I've seen for it still working on Linux was from an AMD user. Something in combination with NVidia maybe?

[Alrik Thingrim]

Hi,
j4yi. I'm running 0.9.4.0 on Linux with AMD and it is not crashing while loading saved games or generating a new world. BUT, the Players.Log is logging tons of errors:
<snip>
Unloading 0 unused Assets to reduce memory usage. Loaded Objects now: 12958.
Total: 12.067000 ms (FindLiveObjects: 2.104000 ms CreateObjectMapping: 0.354000 ms MarkObjects: 9.589000 ms DeleteObjects: 0.020000 ms)
The referenced script on this Behaviour (Game Object 'Block_SmallCockpit_m2_2x3x3') is missing! (Filename: Line: 1775)
The referenced script on this Behaviour (Game Object 'Block_Cockpit_3x3x3') is missing! (Filename: Line: 1775)
The referenced script on this Behaviour (Game Object 'Block_SmallCockpit_2x3x3') is missing! (Filename: Line: 1775)
<snip>

Brief System: {
Ubuntu 16.04 64bit (Linux 4.4.0-116-generic #140-Ubuntu SMP x86_64 x86_64 x86_64 GNU/Linux)
AMD FX(tm)-6300 Six-Core Processor × 6
Desktop is 1680 x 1050 @ 60 Hz
Initialize engine version: 2018.2.4f1 (cb262d9ddeaf)
GfxDevice: creating device client; threaded=1
Renderer: GeForce GTX 750 Ti/PCIe/SSE2
Vendor: NVIDIA Corporation
Version: 4.5.0 NVIDIA 384.130
}

Hope it helps.
BTW, I istalled Version 0.9.5.0 and it is running well on the above system.

Cheers
Alrik

[cthat]

Can confirm.
Version of game 0.9.5.0
OS Linux Mint 18.2 Cinnamon 64-bit
Kernel 4.4.0-137-generic
CPU Intel Core i7-4771 3.50GHz
RAM 16Gb
NVIDIA GK104 [GeForce GTX 770]

Exception Mono.Data.SqliteClient.SqliteExecutionException - SQL logic error during BulkExecute: INSERT INTO [active_blocks]([id], [type_id], [data], [container_id]) VALUES (?, ?, ?, ?)
StackTrace:
at Mono.Data.SqliteClient.SqliteCommand.ExecuteStatem ent (IntPtr pStmt, System.Int32& cols, System.IntPtr& pazValue, System.IntPtr& pazColName) [0x00000] in <filename unknown>:0
at Mono.Data.SqliteClient.SqliteCommand.ExecuteStatem ent (IntPtr pStmt) [0x00000] in <filename unknown>:0
at Mono.Data.SqliteClient.SqliteCommand.ExecuteReader (CommandBehavior behavior, Boolean want_results, System.Int32& rows_affected) [0x00000] in <filename unknown>:0
at Mono.Data.SqliteClient.SqliteCommand.ExecuteNonQue ry () [0x00000] in <filename unknown>:0
at Mono.Data.SqliteClient.SqliteTransaction.Commit () [0x00000] in <filename unknown>:0

[Ehvis]

I did a whole bunch of tests today on the new 0.9.5.1 version. Where I was able to get one to start previously, I wasn't able today. Maybe I was just lucky last time.

I've been experimenting with libsqlite, but I think this may be a dead end. I made sure that it only loaded the one provided (checked the process memory map) and the problems remain the same. The only thing that can be changed is the "Fallback handler ...." errors by creating a symlink to the libsqlite3.so.0 file in the .../Mono/x86_64 directory.

Otherwise the logs are always the same as the one from starting post and the one provided by liamdawe in the Steam forum post. That is, a bunch of "SQL logic error during BulkExecute", an occasional "[SandyEngine #ERROR# DatabaseManager.cpp:81] SQL Erorr: disk I/O error" and if it stays alive long enough, "database disk image is malformed during BulkExecute". Those are for creating a fresh save.

I don't see any things that I can test myself. The only thing I *know* is that it worked fine in March because I triggered a bunch of achievements back then. The one very notable thing that I see in between then and now is the update to Unity 2018. Maybe others can confirm whether or not they got it to run after that.

If not, maybe you can provide a few beta branches with previous version so I can check which update broke the game.

[cthat]

If not, maybe you can provide a few beta branches with previous version so I can check which update broke the game.

I did some research lately. Posted it here. My conclusions, that problem with saves started between 0.9.04 and 0.9.2.2 (or even 0.9.2.1 according to other forum user lucy). All updates starting 0.9.2.x has almost the same problem, including recent update 0.9.5.1. Here's the recent results of attempt to start new game. Saves-121018.zip and Player.log-121018.zip

EDIT: typos

[martinsustek]

Hi guys,

I am really sorry about the Linux issues, I'll do my best to fix it but I am little bit helpless now.


Today, I have tested 0.9.5.1, 0.9.3.1 and 0.9.0.4 on my home Ubuntu machine and was able to run two consecutive survival games in each version without any database error. Even with system sqlite3 package installed or not installed. I've tried builds made on Linux as well as on MacOSX and never experienced that bug.

I wanted to try it in fresh installation of Mint 19 in VirtualBox, but was not able to manage OpenGL Core to work in virtualbox. I'll try it next week with some LiveCDs (or LiveUSB), but that dualbooting takes a lot of time. If you could post here a link to exact .iso distribution that is proven not to work, it would be helpful.



I will summarize some technical facts now, just for clarification - sqlite3 is a third-party database library that is used to store game save data. It is also used during the gameplay to store some temporary data (the state of the current game world that is not currently visible in the scene is in _working.db file). So our save files are actually SQL databases and you can browse (change) them by some third-party tools (and some hackers actually does it .

So, runnng a new game acutally means creating new database file and adding some generated data like alien wrecks and monuments. That's why database is used even during new game creation and not only during save or load. And those bulk inserts are what causes crashes in your case. Without it the world generation cannot continue.

To be able to use those database files, our game must use the library. You probably know, that PlanetNomads consists of two parts, ie. the main game written in Unity (C#, .NET) and the terrain engine Sandy written in high performance multithreaded C++. Both parts are written using different languages and technologies, but both needs to access the database file at the same time.

It needs specially built and configured version of sqlite to be able to access a single database file from multiple threads, otherwise different thread could overwrite its data and corrupt the file. Sandy has sqlite library included and linked statically, you can say built-in, so there should not be a problem, but C# code itself uses a .NET multiplatform dynamic library, that somehow internally loads specific dynamically linked native library for each plaform. And this is where the magic happens. If wrong library is loaded and used for some reason - a version that do not support bulk inserts or multithreaded access, file gets eventually corrupted.

Also there might be some OS-specific differences in handling file sharing issues and some change in new version of Unity in dynamic libraries loading. I am not pretty sure. There are lot of possible causes, and neither is something a game developer want to deal with.

We definitelly do not want to give up on Linux, it will just need some time to find the cause and a solution as it's all Greek to us. Please bear with us.

[Ehvis]

We definitelly do not want to give up on Linux, it will just need some time to find the cause and a solution as it's all Greek to us. Please bear with us.

If I understand correctly, you don't have an install that demonstrates this problem, which makes this rather hard to debug. Is there anything *we* can do to help out? Maybe provide some additional information by enabling more logging. It is no problem for me to do some leg work to help out if possible.