OK... can reproduce this now! Yesterday's lead was false but I have found a way to reproduce this 100% (and I know why it happens). I will not say how it's done at this point! But I'll tell you that the last line of the log file when you got the crash might tell you how to to it.
I'll work on a clean fix and there's a couple of requests to consider. By the way, does anyone use invisible hosts using the /dedicated=invisible option? It doesn't do anything with DCon because the console window just appears - LFS doesn't have any control over that. I'm not sure but maybe you can create an invisible host by piping or redirecting the output...
I'm glad the Sherlock Holmes in you found the culprit so fast Personally, I never really looked into making use of the invisible option.
OFF-TOPIC: I can't resist to ask you for a mini status update of the autocross improvements discussed a few months ago:https://www.lfsforum.net/showt ... php?p=1759364#post1759364 I'm wondering if they are still on your todos list since I'm working on my InSim application anticipating these new features
For this to work you'd have to get the handle of the console window which would require some heuristics and even then it would be error prone (like if you had two consoles with the same title running).
EDIT: You could generate some random string, call SetConsoleTitle() and look for that with FindWindow(), but that sounds like a nasty hack...
For a full GUI-less experience on Windows it would be better to have DCon run as a service.
Encountered a very odd issue 10 minutes ago. Server setup:
Track: AS3
Laps: 3
Cars: FXR only
Players: ~10-15
On the third lap 50m before the leader crossed the finish line the race auto restarted. No SHIFT + R votes were seen. And in my InSim logs no "/restart" was found (my InSim app logs all commands even if ACR UCID == 0).
I tend to think it's a bug so far but I might have missed something very obvious. The only thing I have so far:
Some manifestation of frustration there since the race would have been over (for the top 3) 5 seconds later
Am I right to assume that IS_ACR always reports /restart when it occurs (by vote, admin, InSim app)? Obviously I have no idea how to reproduce it.
Take this report with a grain of salt but if anyone ever encountered this I would gladly hear about it.
EDIT: I may have failed to see the obvious after all. If you look at the logs. The race restarted after the "Repaired Mesh" and especially when "[MRc] Nofreak^L left the pits (FXR)". Most importantly 1 minute before that event: "Jun 15 22:20:03 LOWERSTANDARDS™^L voted to RESTART". Could it be that when someone votes to restart race and afterwards a player spectates the amount of votes required for a race restart drops by 1 thus triggering an auto race restart after all. I suppose I'm close to dismissing this invalid bug report...Will try to reproduce it though.
I've personally seen races restart on a spectate event (caught me by surprise first time!). Indeed I interpreted it to mean that the vote threshold must reduce by 1...
And wth, it's always good to get bug reports so don't feel bad
It seems to be the explanation. I still wonder if this is a bug or intentional. Either way it seems unlikely to be related to TEST5. Finally, what is perhaps worth noting is that "/restart" is not reported with IS_ACR when this auto-restart happens.
If >50% on the track votes for a race restart then it simply restarts yes. So if someone spectates which didn't vote, it makes it easier for the ones who did vote for a restart because it's easier for them to reach a majority.
Two solutions to that;
1) turn off voting
2) use IS_VTC (since 0.6B) to cancel an ongoing vote when it's not the right time to start a vote.
if you mean /vote=no , this has nothing to do with race restart votes. Its for ban/kick votes... so there is no easy way to turn off votes, altought its possible by setting very high /rstmin
1) I don't think you can turn off the ability to SHIFT + R (and SHIFT + X).
2) My InSim app already cancels votes sending TINY_VTC when an IS_VTN is received (one of our admins disabled this InSim feature so in the case I reported no TINY_VTC was sent).
In the end my report intends to obtain some clarity concerning SHIFT + R voted restarts. More precisely if it's normal that a SHIFT + R vote could result in an auto restart later on when someone spectates during the race. I suppose it's not something that was introduced with TEST5. It's because I was taken by surprise I thought it was TEST5 related thus reported it. It should be fairly easy to "reproduce" if you have a bunch of racers on. Suppose 87 players are online and racing. If 3 players vote to restart and at some point 1 racer spectates then an auto restart could be triggered. That is if 50% (or more) of votes are required.
Corrected ;-) Is it a certainty LFS votes work this way? The vote is passed when 50% or more voted? Does this vote condition apply to SHIFT + R, SHIFT + X, kick and ban votes?
Ok good to know although I still wonder what the exact distinction is. Which votetype has which conditions and if there is a difference between players spectating and players in garage. I speculate there is a distinction between spectators vs players on track (or in garage). It's not that hard to test but if someone knows for sure how it works that would be nice.
I think my report should be dismissed as I believe auto-restarts are reported by IS_VTN thus making it possible to cancel the restart if required. Doing "totalvotes required - 1" when someone spectates probably makes more sense anyway, yet it feels far-fetched and confusing but now I understand how it works so I can anticipate things InSim wise
bug
I started my server connected to it. The file deb.log is all recorded and "live progress" reacted to commands /laps 20 or /laps 0. After accidentally starting a second time, it was written "bind failed". After that, the file deb.log nothing new has been written and "live progress" does not respond to commands, and does not change the number of laps.
Thats becuase deb.log is overwritten by other instance of DEDI... Maybe it should somehow lock file and prevent further instances to write... or create deb_2.log
Becuase you cannot start two servers with same /port
I cannot reproduce this
E: BTW, I noticed when you type "//something" (double slash) it always threat that as /run command. I didn't see that anywhere documented...
also from wiki
My server (TEST5) crashed for the 2nd time, tonight, exact same error.
I suppose you won't tell how to prevent partycrashers to cause the server to crash. (Hopefully it's not a stupid move to speculate that it is related to an InSim packet with name starting with IS_M__?:schwitz
Scawen, should we revert to LFS dedi 0.6E temporarily while awaiting TEST6 or do you think TEST6 is close enough to remain on TEST5 for now or perhaps it can still be useful to remain on TEST5 ?
DANIEL-CRO, I think the program should scan processes and make sure that it is in this folder, does not have running processes. And if the program is running, the program does not start and displays a message in a window with the message "the program is running, click OK, to exit."
But I'll tell you that the last line of the log file when you got the crash might tell you how to to it. 
