[INVESTIGATING]Norender server position sync stop - unrecoverable

[discwalker]

Too often and long ago happen in DCS: MP become SinglePlayer! You understand our love-hate relationship with DCS?

(My plane is the starting F-14B)

 

Log:

https://cdn.discordapp.com/attachments/606862310853574667/672509062126239744/dcs.log.stopsync-Aerobaticserver.txt

At the end the last connections can done nothing on server.

 

mission:

https://cdn.discordapp.com/attachments/606862310853574667/672511669099823193/AEROBATICS-20200129-204359.miz

 

On our Aerobatic dedicated server A/C position sync frozen with 22 players.

Only the own plane is moving, everyone else in the air is levitating with zero speed.

After disconnect reconnect was unsuccessful to members, until server is not listed anymore. (died without a crash)

Player list frozen too in MP session, show the disconnected members.

The chat events (landing/takeoff/crash) updated a long time after A/C position sync lost.

This bring back memories from dedicated server multiplayer mass test data collection.

After sync loss go to spectator or change slot is impossible. Only disconnect worked.

The voice chat was off, no SLMOD, no MIST.

The max client ping was 128 ms.

No hardware issue, sure, another two instances run fine on the same physical machine, one with multiple players after the Aerobatic instance become inoperative. (another instances: https://cdn.discordapp.com/attachments/606862310853574667/672511044094263306/Screen_200129_235455anotherinstancesLongAfterAeroDied.jpg)

 

("ERROR Lua::Config: Call error debriefing.addEvent" happenig on our server quiet often too)

[Binary]

We encountered the same today. It is unfortunately a carrier start replay, so probably not worth posting. But here at least the errors from DCS.log.

 

Edit: I've removed the logs since they are related to a script in the mission and since I've checked the tacview and there was nothing in the logs that aligns to the time the position updates stopped working.

 

Another sideeffect, besides no position updates anymore, were horrible stutters (1 sec stutters every about 5 seconds).

Edited February 15, 2020 by n26

[robgraham]

When this happens go look at the server, you'll find that the actual server instance is non-responsive to windows and that the log output is slowed down and coming in bursts etc.

 

It's not something 'new' though it's become more touchy, it's why a lot of servers are running time out monitoring on the instances, so that if it goes for an extended period of time the server instance is killed and restarted.

 

The biggest items that tell me that is what happened though is your discription of

1. Hung air craft while every ones 'personal' aircraft keeps moving.

2. not being able to select a slot once back on choose slot.

 

if you 'wait' long enough things will move, in fits and jerks, as the process catches up and you might even be able to briefly get into a slot etc. I agree though it's annoying, it's frustrating as hell and until eagle starts doing more MP testing for more then short periods I don't know if we'll ever see it fixed.

 

The only 'fix' most of us have is to restart the server instance.

[BIGNEWY]

Hi

 

I think this is related to multicrew, if the co pilot disconnects while in control.

 

We are investigating.

 

thanks

Edited February 16, 2020 by BIGNEWY

[Binary]

Hi

 

I think this is related to multicrew, if the co pilot disconnects while in control.

 

We are investigating.

 

thanks

 

Thanks for looking into it. It might be that the initial post of the thread was a different issue. Since we ever only encountered it since the 2.5.6, and in our case there was no multi-crew in use.

 

It also just happened again. Same effect:

- No position updates from server anymore

- Constant stutter

 

Both start at the same time.

 

The first time it happened, the stutter went away once some players had disconnected (though position updates weren't restored). The disconnecting players were all on the carrier which is why we tried the following this time. We had the one machine on the carrier take-off from it. Once it took off, the game of the player of that machine froze, but the stutter for the other players was gone.

 

There was again nothing strange in the logs for the time the issue started and the server CPU and RAM usage during the stutter was normal. I have both the replay and the tacview, but the issue happend 1:08h after the start. So they are quite long which is why I haven't attached them for now. If they are despite their length still helpful, feel free to let me know and I'll look for a way to attach those 100 MB.

 

EDIT: I've checked my server again and must have missed that the server (in norender) ate all my 32GB of RAM (it usually only takes around 4GB), Just in case that this might be related and helpful to know.

EDIT2: Unable to reproduce by simply sitting on the carrier and letting the mission run for a while

EDIT3: Just noticed from watching the Tacview files again that the position updates only stopped for aircrafts, but not for ships

Edited February 17, 2020 by n26

[Wiggy]

I have also noticed this same issue testing missions on my no-render dedicated server machine. I had previously thought it down to my server PC not being powerful enough as it’s a heavy mission, however, last time this happened, and as you say, the aircraft (all AI except myself flying) froze position, I could still cause damage and the flak was still firing on them. The rest of the mission was still functioning - artillery was firing, flak still firing, ground units still active. The only units which froze up were the aircraft. After flying around and waiting they occasionally and randomly switched positions and ‘updated’. One occasion it’s actually fixed itself and the battle continued without another freeze.

 

This mission uses exclusively WW2 units by the way.

 

I checked the status of the server during the freezes and it was intermittently ‘not responding’, but always came back and never actually fully crashed.

 

I’ll do some more testing and try and reproduce it. For me this is encouraging in a way as it may not mean my machine

isn’t up to it:)

 

Edit: also to note this was occurring before the 2.5.6 update as well as after.

Edited February 19, 2020 by Wiggy

[BIGNEWY]

if you have logs and a track replay that is reasonably short please attach them, it will help.

 

thank you