Crash ( test build (build 2022-01-04.0a71f9c) )

Describe your experience with the latest version of FreeOrion to help us improve it.

Moderator: Oberlus

Forum rules
Always mention the exact version of FreeOrion you are testing.

When reporting an issue regarding the AI, if possible provide the relevant AI log file and a save game file that demonstrates the issue.
Message
Author
wobbly
Cosmic Dragon
Posts: 1879
Joined: Thu Oct 10, 2013 6:48 pm

Crash ( test build (build 2022-01-04.0a71f9c) )

#1 Post by wobbly »

Vezzra's windows build.
test build (build 2022-01-04.0a71f9c)

Steps to produce:

Start a game
Wait a couple of minutes
Game exits to desktop

If I'm doing something while it happens game will freeze up and become non-responsive
Attachments
freeorion.zip
(15.03 KiB) Downloaded 47 times
Last edited by wobbly on Thu Jan 06, 2022 4:16 am, edited 1 time in total.

User avatar
Geoff the Medio
Programming, Design, Admin
Posts: 13587
Joined: Wed Oct 08, 2003 1:33 am
Location: Munich

Re: Crash ( test build (build 2022-01-04.0a71f9c) )

#2 Post by Geoff the Medio »

I'm not able to reproduce this. Maybe there is something interesting at the end of one of the log files after it happens?

wobbly
Cosmic Dragon
Posts: 1879
Joined: Thu Oct 10, 2013 6:48 pm

Re: Crash ( test build (build 2022-01-04.0a71f9c) )

#3 Post by wobbly »

Code: Select all

02:33:30.348031 {0x00000b30} [debug] FSM : serverfsm.cpp:305 : ServerFSM::HandleNonLobbyDisconnection : Lost connection to player #1, named "Human_Player".
02:33:30.348031 {0x00000b30} [warn] FSM : serverfsm.cpp:335 : Player #1, named "Human_Player"quit before empire was eliminated.
02:33:30.348031 {0x00000b30} [debug] python : <string>:3 : Python current directory set to C:\Program Files (x86)\FreeOrion\default\python\chat
02:33:30.348031 {0x00000b30} [info] python : chat.py:41 : Chat : [[PLAYER_LEFT_GAME,Human_Player]] (255, 255, 255, 255)
02:33:30.348031 {0x00000b30} [error] FSM : serverfsm.cpp:393 : Too high unconnected human player 1 expected 1; server terminating.
02:33:30.348031 {0x00000b30} [error] FSM : serverfsm.cpp:402 : Unable to recover server terminating.
02:33:30.348031 {0x00000b30} [debug] FSM : serverfsm.cpp:3080 : WaitingForTurnEnd time: 141 s
02:33:30.348031 {0x00000b30} [debug] FSM : serverfsm.cpp:2565 : PlayingGame time: 141 s
02:33:30.348031 {0x00000b30} [debug] FSM : serverfsm.cpp:3506 : ShuttingDownServer informing AIs game is ending
02:33:30.348031 {0x00000b30} [debug] FSM : serverfsm.cpp:3519 : ShuttingDownServer expecting 6 AIs to ACK shutdown.
02:33:30.373513 {0x00000b30} [debug] FSM : serverfsm.cpp:3540 : Shutdown ACK received for AI 4
02:33:30.373513 {0x00000b30} [debug] FSM : serverfsm.cpp:3540 : Shutdown ACK received for AI 7
02:33:30.373513 {0x00000b30} [debug] FSM : serverfsm.cpp:3540 : Shutdown ACK received for AI 6
02:33:30.373513 {0x00000b30} [debug] FSM : serverfsm.cpp:3540 : Shutdown ACK received for AI 3
02:33:30.373513 {0x00000b30} [debug] FSM : serverfsm.cpp:3540 : Shutdown ACK received for AI 2
02:33:30.373513 {0x00000b30} [error] FSM : serverfsm.cpp:286 : A PlayerChat event was not handled by any of these states : [struct ShuttingDownServer].  It is being ignored.
02:33:30.373513 {0x00000b30} [debug] FSM : serverfsm.cpp:3540 : Shutdown ACK received for AI 5
02:33:30.373513 {0x00000b30} [debug] FSM : serverfsm.cpp:3574 : All 6 AIs acknowledged shutdown request.
02:33:30.373513 {0x00000b30} [debug] server : serverapp.cpp:153 : ServerApp::~ServerApp
02:33:30.373513 {0x00000b30} [debug] server : serverapp.cpp:160 : Server exited cleanly.
02:33:30.373513 {0x00000b30} [debug] server : pythonbase.cpp:99 : Cleaned up FreeOrion Python interface
02:33:30.389139 {0x00000b30} [error] network : servernetworking.cpp:208 : PlayerConnection::~PlayerConnection: shutdown error #10009 "The file handle supplied is not valid" for player id 5
02:33:30.389139 {0x00000b30} [debug] server : dmain.cpp:124 : freeorion server main exited cleanly.
Attachments
freeorion.zip
(15.03 KiB) Downloaded 51 times

wobbly
Cosmic Dragon
Posts: 1879
Joined: Thu Oct 10, 2013 6:48 pm

Re: Crash ( test build (build 2022-01-04.0a71f9c) )

#4 Post by wobbly »

Tested with firefox closed and it seems to take longer to occur but still exits. Hard drive has plenty of space and I'm not seeing a spike in memory or cpu usage.

User avatar
Geoff the Medio
Programming, Design, Admin
Posts: 13587
Joined: Wed Oct 08, 2003 1:33 am
Location: Munich

Re: Crash ( test build (build 2022-01-04.0a71f9c) )

#5 Post by Geoff the Medio »

Those logs have inconsistent timestamps (02:33:30 for server and 02:38:26 for the client). Was the client exit when you were just on the intro screen, not in a game? But the server log from a previous sudden exit while in a game?

Was the FreeOrion window visible and had input focus or was it in the background while you did other stuff?

Can you try with more logging enabled?
options window log options page
options window log options page
log_options.png (115.71 KiB) Viewed 1781 times

wobbly
Cosmic Dragon
Posts: 1879
Joined: Thu Oct 10, 2013 6:48 pm

Re: Crash ( test build (build 2022-01-04.0a71f9c) )

#6 Post by wobbly »

fresh install. Run freeorion (at 21:03). Changed logging options to match your screenshot. Single players (all default settings). Left it running with window visible and focused.

Log is over a million lines. I had to split the zip and then rename the extention (.001.zip instead of .zip.001) for the forum to accept it
Attachments
freeorion.002.zip
(1.9 MiB) Downloaded 59 times
freeorion.001.zip
(4 MiB) Downloaded 51 times

User avatar
Geoff the Medio
Programming, Design, Admin
Posts: 13587
Joined: Wed Oct 08, 2003 1:33 am
Location: Munich

Re: Crash ( test build (build 2022-01-04.0a71f9c) )

#7 Post by Geoff the Medio »

The client log ends at 21:04:45 and the server log has a line at 21:04:45 and then nothing until 21:08:55 when it loses the connection to player #1:

Code: Select all

21:04:45.508079 {0x000025d0} [debug] server : serverapp.cpp:2186 :  ... have orders from empire id: 7
21:08:55.334955 {0x000025d0} [trace] network : servernetworking.cpp:870 : ServerNetworking::DisconnectImpl : disconnecting player 1
21:08:55.334955 {0x000025d0} [debug] FSM : serverfsm.cpp:305 : ServerFSM::HandleNonLobbyDisconnection : Lost connection to player #1, named "Human_Player".
21:08:55.335951 {0x000025d0} [warn] FSM : serverfsm.cpp:335 : Player #1, named "Human_Player"quit before empire was eliminated.
I wonder if your network driver has some sort of timout that's disconnecting after too much idle time?

Could you try sending a chat message every 1 min or so, and see if that keeps it from disconnecting? Or just ending turn every min or so?

wobbly
Cosmic Dragon
Posts: 1879
Joined: Thu Oct 10, 2013 6:48 pm

Re: Crash ( test build (build 2022-01-04.0a71f9c) )

#8 Post by wobbly »

Geoff the Medio wrote: Thu Jan 06, 2022 11:09 am Could you try sending a chat message every 1 min or so, and see if that keeps it from disconnecting? Or just ending turn every min or so?
Neither of these worked

User avatar
Geoff the Medio
Programming, Design, Admin
Posts: 13587
Joined: Wed Oct 08, 2003 1:33 am
Location: Munich

Re: Crash ( test build (build 2022-01-04.0a71f9c) )

#9 Post by Geoff the Medio »

I'll try adding more logging detail, but I'm not sure what else to try or test with the latest version.

You could try some older test releases to see if there is one in particular where it starts happening...

wobbly
Cosmic Dragon
Posts: 1879
Joined: Thu Oct 10, 2013 6:48 pm

Re: Crash ( test build (build 2022-01-04.0a71f9c) )

#10 Post by wobbly »

Geoff the Medio wrote: Thu Jan 06, 2022 12:10 pm I'll try adding more logging detail, but I'm not sure what else to try or test with the latest version.

You could try some older test releases to see if there is one in particular where it starts happening...
The previous test release worked for me. I can probably narrow it down to the commit if that would be helpful? Is that just "browse the repository at this point in the history", download, compile? I'll have to wait till I have more time.

User avatar
Geoff the Medio
Programming, Design, Admin
Posts: 13587
Joined: Wed Oct 08, 2003 1:33 am
Location: Munich

Re: Crash ( test build (build 2022-01-04.0a71f9c) )

#11 Post by Geoff the Medio »

wobbly wrote: Thu Jan 06, 2022 12:17 pmI can probably narrow it down to the commit if that would be helpful?
yes.
Is that just "browse the repository at this point in the history", download, compile? I'll have to wait till I have more time.
If you don't want to use git to jump between different commits quickly, then that would be a plausible workaround.

Ophiuchus
Programmer
Posts: 3433
Joined: Tue Sep 30, 2014 10:01 am
Location: Wall IV

Re: Crash ( test build (build 2022-01-04.0a71f9c) )

#12 Post by Ophiuchus »

wobbly wrote: Thu Jan 06, 2022 12:17 pmIs that just "browse the repository at this point in the history", download, compile? I'll have to wait till I have more time.
yes, also you can git reset --hard to a specific revision.

And also helpful but maybe a bit confusing at the beginning there is support for bisection in git:
a way to take the middle between two commits and can check if it is working and dependending on that take either the newer or the older half as input for the next bisection step until you found the last working version.
Any code or patches in anything posted here is released under the CC and GPL licences in use for the FO project.

Look, ma... four combat bouts!

wobbly
Cosmic Dragon
Posts: 1879
Joined: Thu Oct 10, 2013 6:48 pm

Re: Crash ( test build (build 2022-01-04.0a71f9c) )

#13 Post by wobbly »

[build 2021-12-31.63157c7] MSVC 2022 - working (at least based on a 20 minute test)
[cdad629] - crashes

I can test the builds in between when I have more time, I think I've had enough of endlessly compiling FO for 1 day.

Ophiuchus
Programmer
Posts: 3433
Joined: Tue Sep 30, 2014 10:01 am
Location: Wall IV

Re: Crash ( test build (build 2022-01-04.0a71f9c) )

#14 Post by Ophiuchus »

wobbly wrote: Sun Jan 09, 2022 7:32 pm [build 2021-12-31.63157c7] MSVC 2022 - working (at least based on a 20 minute test)
[cdad629] - crashes

I can test the builds in between when I have more time, I think I've had enough of endlessly compiling FO for 1 day.
based on the behaviour (no interaction with the game) i guess the follwing commit could be the culprit?

is this already bad?:
commit fb38731469610571e01bbda674bad396be797bc5
-de-pointerize RenderCache
-don't redundantly create RenderState objects when one was passed into Font::PreRenderText

and this one still good?
commit ff62820ae17d425b7e9fabd25981333f500b3db5
reserve slightly more space in vertex buffer, which was often one smaller than required
Any code or patches in anything posted here is released under the CC and GPL licences in use for the FO project.

Look, ma... four combat bouts!

wobbly
Cosmic Dragon
Posts: 1879
Joined: Thu Oct 10, 2013 6:48 pm

Re: Crash ( test build (build 2022-01-04.0a71f9c) )

#15 Post by wobbly »

Ophiuchus wrote: Sun Jan 09, 2022 10:02 pm based on the behaviour (no interaction with the game) i guess the follwing commit could be the culprit?

is this already bad?:
commit fb38731469610571e01bbda674bad396be797bc5
-de-pointerize RenderCache
-don't redundantly create RenderState objects when one was passed into Font::PreRenderText

and this one still good?
commit ff62820ae17d425b7e9fabd25981333f500b3db5
reserve slightly more space in vertex buffer, which was often one smaller than required
Yes this is correct.

Post Reply