[01.197.170] re-connecting-to-server-after-character-has-been-removed-causes-crashfreeze

Nikolas March shared this bug 7 months ago
Need More Information

recreated ticket with more info

various crashes connecting to servers that have to re-create (or reconnect to) the player's character (like keen officials)

every time i reconnect to one ive been playing on (keen uk #5) i get a crash sometimes with and sometimes without a crash reporter popup, or this freeze resulting in unable to use my PC (see attached image)


i have also sent a few crash reports with the crash reporter when it popped up

i (was) on that server as "Marcus" (7656*********8661)

edit: another 3 times to connect, the logs from the 2 failed attempts included again


55 days pass by, then...


this is happening all over again, but on keen uk 2 (on as Marcus and Nikolas March, 7656*********2485), its also locking steam up, i had to end steam from task manager


3 hard crashes, then a successful connection to keen uk 2...


someone having the same problem with a worse success rate..


keen discord #help-se

https://discord.com/channels/125011928711036928/136097351134740480/810642236487499787


updating graphics drivers and deleting shadercache2 did not help

multiple attempts, used DDU, driver reinstalls, still CTD...

my 2nd pc and 2nd account can connect fine to uk 2...

and look at that, my 2nd account there, woke the area up as it was in the same area and now my main pc main AND account can connect. nothing wrong with my PC or drivers!

while my second account was on, i toggled the spawn points, this also may have helped my main get back in...

there is also a 'mipMapTextureCache2.B5' file in the same folder as the logs with the same date/time, ive moved it elsewhere in case you want it, 16MB


dropbox links for dumps using VS debugger, nvidia driver fail on my main PC


Exception thrown at 0x00007FFB6743F8FF (nvwgf2umx.dll) in SpaceEngineers.exe: 0xC0000005: Access violation reading location 0x000000000000010C. occurred

minidump without heap, 3.6MB

https://www.dropbox.com/s/id9r23zvg4cxi5f/SpaceEngineers.dmp?dl=0

minidump with heap, 9+GB

https://www.dropbox.com/s/0qltg7nr2uzdmnp/SpaceEngineersHeap.dmp?dl=0


not sure if drop box links will work, im beyond my storage limit which is 2GB

Google Drive minidumps with VS debugger from my second PC

Exception thrown at 0x000007FE702C3B59 in SpaceEngineers.exe: 0xC0000005: Access violation reading location 0x0000000000000008

minidump without heap 2nd PC 3.7MB

https://drive.google.com/file/d/1KspTfIHK01t4OT6VPcwwY4OoneBBMmEc/view?usp=sharing

minidump with heap 2nd PC 8.2GB

https://drive.google.com/file/d/1U9b5r5WpLGxwJ8thh9rox0xvu9xZexWU/view?usp=sharing


PC specs

Main, 5930k OCd to 4.5GHz, 32GB 2400MHz DDR4 quad channel, GTX980 stock clock and power limits, windows 8.1 home, main drive 3x SSD raid 0

2nd, 3930k slightly OCd to 3.8GHz, (i usually run at 4.3GHz, but there is no power saving mode so it runs hot all the time at that speed) 16GB DDR3 1866Mhz quad channel, GTX970 stock clocks and power limits, windows 7 professional, main drive single SSD

both CPU overclocks maximums were stress tested and tweaked accordingly over many days for each PC, this was 2 years ago.

Comments (2)

photo
1

title slightly incorrect, can also happen while your character is still there...

photo
1

Hello, Nikolas!


Thank you for your detailed report. I tried to reproduce your issue on UK #2 server but so far without any luck. I tried these reproductions steps:

  1. connect to a UK #2 server
  2. have a powered respawn point
  3. log out and wait till a server removes my character
  4. connect again

I see you've already tried to update your graphics drivers and delete your Shadercache folder. Could you please try to back up and delete your whole Space Engineers roaming folder and verify your game files in Steam? We will appreciate any additional information. Thank you.


Kind Regards

Keen Software House: QA Department

photo
1

have you tried doing it on a server where it does happen? like keen uk 2, start in space with a spawn pod, use spacemaster tools to place it near an asteroid, and to place merge blocks and merge to asteroid to make it permanent, take blueprint of it, somewhere near earthlike.

then do other things that players would do, fly around, in and out of sync distance, drill some asteroids... then log of (with admin/spacemaster tools disabled of course)


after deleting shadercache again, and deleting the main menu videos, i had 2 more CTD, then was able to connect, that was 15 or 16 in total yesterday (thursday 18th), im expecting a similar situation when i next try to connect.

photo
1

i skipped playing most of saturday and all of friday, was able to log in without issues late saturday, played a few hours and then sunday morning just before and after the restart, logged off for a break and a snack, tried to connect 20 minutes after log off, now im back to the CTDs, and again there is this new file that has been updated at the time of the first failed reconnect attempt, "mipMapTextureCache2.B5", in the same folder as the logs are placed

google drive link

https://drive.google.com/file/d/1udMe62MzIuQaJQQ6DrjR51YZQExY423b/view?usp=sharing


after deleting all of those mipmaptexturecache files, and folders shadercache and shadecache,, i had another CTD while connecting to keen uk2, i then tried keen uk 5, connected and loaded fine, it gave me the respawn ship menu as i havent been there in a while... tried keen uk 2, got the crash reporter with the generic graphics driver problem message on 2 connection attempts, then back to CTDing on each attempt.

photo
1

continual CTD with main pc and main steam account, so i tried my main account with my second PC, it was able to connect once without problem, then subsequent connections resulted in CTD... and switching account on second PC back to second account did not help... got a warning sound similar to 'unable to place block' then generic graphics driver crash message.

photo
1

restaring both computers, main PC and account successfully connected, i saw the second account character there, he died while the second PC was failing to connect in, the second account SE on second PC-> CTD

tried agaion on the second PC, i saw the character spawn from the viewpoint of my main account and main PC, then SE on the second PC-> crashed with reporter... generic graphics card driver message

photo
1

i played all afternoon on keen uk 2, took a 45 minute break for dinner, now i CTD trying to connect...

used batch file to delete mipmaptexturecache files, and the two folders shadercache and shadecache2, and now im in... this didnt work immediately when i did it earlier, this is the most inconsistent indiscriminate bug ive seen. im glad im not a game dev.

photo
1

Hello, Nikolas!


Thank you for your replies, I'm struggling to reproduce it. I tried to connect to UK #2 several times without any issues. I would suggest you to back up and delete the whole Space Engineers roaming folder, not just shadercache files etc. Also a game reinstall could help. We will continue the investigation.


Kind Regards

Keen Software House: QA Department

photo
1

since i have been deleting those 2 folders and the mipmaptexturecache files i have had more than 75% success rate getting on the server

after the first time doing this, i did spot some odd graphical behaviour after managing to get in, this cleared when i switched to medium models and then back to high, i wasnt able to record it as i had just restarted the PC and forgot to start OBS recorder.


the best thing to do, would be for me to get one of my characters in that situation that i cant connect, and then instruct you on the grid name, for you to take control of the grid, and then get respawning at it with spacemaster turned off... either by using spacemaster to join my faction would be the best way, as to not disturb the grid itself by changing ownerships

photo
1

the removal of SE videos, and deletion of those 2 shadercache folders and the mipmaptexturecache files didnt work on my second PC, until i restored nvidia control panel Global and SE's 3D settings back to default, i had also done this on my main PC but that alone didnt give any positive results until i had also removed the SE videos and started regular deletion of those 2 shadercache folders and the mipmaptexturecache files...

i have a feeling this bug wont be fixed, and best we can do is get a good work around, which itself may help figure out how and why this happens.

photo
1

i use these 3 lines in a batch file and run it when SE crashes


rd /s /q %appdata%\Spaceengineers\ShaderCache

rd /s /q %appdata%\Spaceengineers\ShaderCache2

del /q %appdata%\Spaceengineers\mipMapTextureCache*.*

photo
1

Hello!


We managed to reproduce this issue following your steps with merging space pod to asteroid. Unfortunately it was only possible to reproduce it a few times. We will continue with investigation. We think that it could be helpful if you contact the server admins on UK #2. They could check your grid and remove your character / the merge block or other parts of a grid. We already dealt with a similar issue where the merge block was turned off and deleting it solved the problem. Could you please try it?


Thank you.


Kind Regards

Keen Software House: QA Department

photo
1

there were no merge blocks owned by me for the entire time i was having the problems...

the grid that i owned that was 'merged' to remove the temporary tag, is one of the ships i have now, but i have had issues logging in, to spawn at grids that werent previously 'merged' in anyway, i do have a grid with merge blocks on it, but that has only been recently built, way after these problems were occuring...

the issue for me has lessened since i have been doing the workarounds to delete those 2 folders and the mipmaptexturecache files when the CTD happens, i can then get in either straight away, or after one more CTD (and occasionally just two more CTDs)


someone cleverer than me suggests you look into LCD sprites of nearby cockpits/flight seats, as deleting those mipmaptexturecache files was the 'change' that reduced the CTD from 9/10, to 3/10

i do have a set of them saved, i can add them to google drive if you want them (sprite data is in those files)

photo
1

ive set the google drive links to the right access, i thought i had already done this, so sorry :)

photo
1

update, problem still persists, i had one occurrence of having 7-8 crashes before being able to get on despite using the workaround, this is the first time i wasnt as successful as before, the only difference than normal was that the server was nearly full at the time, as i was fighting for the last player slot, but later on that day i was back to the normal, 50% chance of a Crash with only 1-2 crashes after i do the work around.

photo
1

This keeps happening to me as well except the game doesn't crash exactly, it just gets stuck on the User host has disconnected... Reconnecting in progress window. After 20 mins I get tired of waiting and go with cancel. Keen#2. It's impossible to get a game going because ur pod gets removed before you can do much. Reconnecting on server restart is broken for me. This function did work at one time but no more.

photo