Issue with peds spawning in player cars

This is what i have figured out so far.

Running 2766
We had this issues with peds just falling out of cars including air craft and blimps at a very high rate. So we had car pile ups every where and blimps falling from the sky every where.

Also peds that spawned in for our drug mission who are normally aggressive and combative. About 25% of them would stand their stupid or run away.

Running 2430
Fixed the issues but lost the tx admin up date which we really want.

Running 2641.
Brought the issues with peds spawning into cars randomly when you dont occupy them
and also brought back the non aggressive ped issue but fixed the peds falling out aircraft.

Im going to try 2569 tomorrow since thats when the txadmin update was released. Hopefully
its before the ped issues started.

How about instead of blindly trying random versions you provide steps to cause these ‘ped issues’ instead?

There’s various servers running on newest versions even with 400+ slots and AI enabled that don’t have this issue so it’s got to be related to something you’ve set in configuration or a script you’re using.

Ok step 1 get out of your vehicle. Step 2 get mad because ped is in your car.
or version two
Step 1 Dealer spawns in a car at the dealer to show someone. Step 2 get mad because a ped is in the car.
Like other have said there is no rhyme or reason to it.

Also the latest version for us doesnt have the peds stealing cars or spawning in them. Just falling out of their own cars/aircraft every where. Also some of them being dormant or or running from aggressive situations in scripts where natives typically have them being aggressive.

We are not blindly trying versions. If we can figure what update caused stuff to stop working we can
probably figure out where the conflict is.

Yeah most servers atm are running old server artifacts!
In our Dutch category most of them are running on older artifacts.
Until now i have not seen any server running the latest artifacts and
having no issues at all. Also the most people that start a server are not developers.
So they probably will take artifact version 2430 as this is the “Latest Recommended”
standing at the top with a big nice blue button. And if you call something
the “Latest Recommended”, you can safely assume most of them will
choose that version!!.

So if you say there are @nta
please show us a few that do, because for all i see right now is those that
did upgrade have weird OneSync issues at the moment.

Saying immediately that it probably lays in one of the servers resources is bogus.
For us personally we have not changed or edited any resource files at all lately, and
also we have the same issues as these other servers. We have not 1 single error in the
client or server console. Everything was just running fine until FiveM descided to update OneSync not all to long ago, thats where all these issues started to happen.

So again none of our resources has been changed, removed or edited. Our server was running just fine. Then FiveM has some update and now everything is buggy has hell (with the AI).

  • People not able to see a spawned in car for some time.
  • Traffic that person A sees, person B does not see and drives right through it.
  • AI randomly spawning behind the wheel of the players car on exiting.
  • AI randomly spawning on the map being and staying static
  • Cars randomly spawning in on top of each other, blowing it all up after.
  • Players invisible and unable to see each other.
  • AI Cars often are empty, no driver visible yet the car is driving.

The above summed problems are OneSync issues, things we are unable to fix server side.
Furthermore if we revert to a older artifact version the issues stop to exist, what is confirming that the issue lays within the artifacts and not with one of our resources. On top of that if we run a clean server with just the server files, we have the same issues!!! So if the issue presists without ANY plugins installed. Explain to me how this is a resource issue ? @nta

Also your saying to @Degeneratepos not to blindly trying all kind of server version, but why ?
If you want to debug a problem you a stupid NOT to try different versions of the server Artifacts. If you are not trying other version to see the results, you wound be able to debug something in the first place. So its actually good he is trying to debunk this with different versions.

Bottom of the story. FiveM did a update - then we experience a lot of issues - we request help and like always, FiveM points the finger to us and tells the problem lays with us. FiveM grabs over 100000 dollars per month via patreon for INSANE license prices, wile at the same time they have their software open sourced and are unable to push updates without introducing many bugs each time. I have spoken to one of the devs once on the CFx.re Discord about problems we experience before these issues. After some talk he said to me “You are always welcome to help us our with development, you can find the repo on github”. But here lays my biggest issue. If your program is open sourced and you do depend on outside developers, then why are you guys asking that much money per month for 128+ players. What the heck are you guys doing with 100k per month ?

If you guys would use that money you so get from all of us each month to actually hire some good developers instead of wasting your time with introducing a complete new Client UI, or those stupid jokes on starting the client, or the music intro you lately added, or the old GTA layout you get on start up. You guys lately are wasting your time to change the looks and feels of the client while we are getting drowned in crashes and client issues.

I think you guys need to prioritize your workflow once and fix the client, instead of adding in all these extra things that are useless !!! And hire some good developers for once instead of depending on other developers do on your github repository. And you guys know your system is buggy as hell as you guys are making jokes about it in your client intro slogans what looks so freaking unprofessional. From a developers standpoint, if i was new to FiveM and saw those slogans, i would do a 180 and walk away. Because if FiveM itself is not talking them selfs and their own software serious, how can you expect developers taking it serious. ? no one like to work with a system that feels fundamentally broke!

And as above, this is not some inherent inevitable issue in the latest versions. Read what I said above.

So? That doesn’t mean nobody else runs a newer version.

At the very least:

  • RSM Freeroam
  • GTALife FR
  • FDG Australia

Each of these updated and confirmed no such issues occurring whatsoever - which made us assume the latest versions as known-good.

I’m not saying such. I’m only saying the issue might be exacerbated by certain resources or configuration settings, for it does not seem to be reproducible or inevitable.

“decided”

You are aware this is not some unilateral decision, and we don’t intentionally push broken updates on you, right?

Users wanted improvements and tweaks, and I have been struggling for weeks to get these to work reliably, until the people I had test stuff reported the new versions worked reliably.

And do you have any reproduction steps for these issues, given how neither I, the aforementioned server owners, and half of #releng on Discord seem able to reproduce any new behavior issues at all, and some even claimed that while they did occur at some point, a large amount of them were fixed subsequently?

Also, many of these issues have been reported since forever and have nothing to do with new changes nor are inevitable again.

How are you ‘unable to fix them’ server side?

FiveM, including OneSync, is open-source on GitHub: https://github.com/citizenfx/fivem

Since the issues occur for you, you can debug them yourself and fix them for everyone.

They don’t occur for me or for anyone else I’ve asked to test.

Sure, if you explain to me why these issues don’t occur for me, or any other folks we’ve asked to test ‘without ANY plugins installed’.

Also, in my post I said ‘or configuration’.

That includes:

  • Console variable settings.
  • Network connectivity on any hop between end user and server.
  • Server OS, OS version and specifications.
  • Player systems, player positions, player count, player behavior, etc.
  • etc., many of these not being scripts.

Again, combination-induced issues do exist. Nowhere in this topic have I implied that a regression is solely induced by user resources.

In fact, if I were so convinced of that, I’d not have replied to this topic at all as I’d not have been trying to find out what is causing these issues, for they do not occur for me or anyone’s servers which I have access to.

I’m merely trying to help figure out the underlying cause and fix these issues for those people having them. Without them happening to myself, this is impossible.

He wasn’t trying ‘all kinds’ of versions, he was making random jumps between various unrelated versions.

In addition to that, oftentimes a build contains multiple commits, and if trying to find the root cause of some issue, going build-by-build is the primary step.

Plus, it’s already known when these issues started occurring for most and when they worsened/improved again for some.

It, however, is not known where in the code these issues originate, due to the complex interaction of client/server code with Rockstar’s game logic, and these issues not occurring for myself or any other scenario I have access to.

Continued in a second reply.

It is however misguided to just randomly pick versions instead of trying to ascertain a starting point, and it’s already known which bug fixes (not arbitrary changes or ‘improvements’) led to these issues happening for some and a 2-week sprint being entirely dedicated to getting to the bottom of these issues until server owners we’ve asked to test confirmed they did not occur anymore in their case.

Them still doing for some others, in fact, is exactly why we didn’t blindly update the ‘Recommended’ build tag. It’s also exactly why I’ve been replying to these topics, trying to get information from the users that experience these issues.

I’m not acting as if they don’t exist, I’m just stating they’re not reproducible for me or anyone I know, so it’s not possible to fix them that way.

No.

Nowhere am I doing so at all.

Nowhere have I done so at all.

If I believe an issue is with the user, I’d not be replying to a topic at all, but letting others do so only.

If I reply to a topic, I am providing a good-faith request for more information to help reproduce the issue.

Part of the CitizenFX philosophy is that there is no user error.

If any user can do something wrong, that’s a flaw in the framework itself.

That does not take away that certain issues only get exposed with specific user code or configuration, that is often not directly accessible to me, so I can’t fix it until someone provides a magic hint.

Wild speculation and blatant misinformation.

Server updates pushed to the ‘latest’ list are not even meant to be bug-free; that’s what the ‘Recommended’ and less so ‘Optional’ labels are for.

Nobody can write software without bugs the first try around.

Client updates get pushed as soon as there’s a known-good state for over 99% of users. That certain edge-case systems experience issues is beyond our control as there are millions of user hardware/software/network configurations possible.

Again, wild speculation.

In reality, any amount will lay much lower due to 60% tax pressure, server costs, additional tax pressure from VAT, software licensing and legal fund contributions.

Your hypothetical ‘100k’ by that point would have less than 10k left a month to actually spend on anything.

We have tried this, however nobody in the community seems to have the highly specialized 0.001% skills one needs to be able to work on core issues related to this project.

It’s not like people with this very specific reverse engineering skillset are just ripe for the picking.

Remember this topic?

Nobody with any proven skills applied at all.

Huh?

You are aware intro UI has zero relation to in-game issues, and is not even taking up any time spent on the latter?

In-game issues are generally either fixed immediately upon discovery, or if that does not happen, it is because they are not reproducible in a guaranteed way, and there’s no live-trace info or full memory dump provided by the people who do have those issues.

You’ll see countless examples on the forums of people posting a repro, or a full dump, or similar, and having their issues fixed within a week from doing so.

Similarly, the examples of issues you see not fixed are generally those where they are generally unreproducible, and no users have provided any tracing or post-mortem information on the problem.

This isn’t some discrete software system where everything is developed in-house, it’s a modification to an entirely undocumented video game which has infinite possible unknown interactions.

It’s perhaps even a miracle it works so well for a very large amount of users.

Again, discounting the lack of correlation between UI design and actual in-game issues, if you’re having any ‘crashes and client issues’, report them. Provide (full) memory dumps. Provide the resources you were running at the time of a crash. Provide the assets you were streaming, and the positions players were in. Provide or write logs that track down entity state whenever some replication issue occurs.

There’s over 10,000 different servers, there are over 5 million unique client systems, 50,000+ unique resources - even an automated system given normal amounts of time can’t test them all, let alone a small set of humans, and yet each of them runs a chance of leading to an unforeseen interaction with any of the infinite variables at play.

That’s why we rely on end users to report issues they see, server owners to relay issues rather than long-form untargeted complaints, and metadata reporting on crashes shows that most top issues don’t have enough information available in automatic reporting, yet nobody is rising up to send their resources, their reports, their full memory dumps, etc. for many of these issues.

Again, what ‘extra things’?

In recent months, many things have been added that actually got fairly good reception.

Also, a crazy amount of top crashers got fixed based on user reports.

That some issues still persist for some implies that the people experiencing those issues have not provided a clear repro for them.

…?

Random jokes based on community reactions have zero things to do with an admission of guilt.

FiveM works fine for the vast majority of users.

We are aware it doesn’t work fine for a small amount, and nowhere are we saying this isn’t the case.

As I’ve said many times above, I can’t fix any issue that I can not induce or analyze otherwise.

It’s like you’re asking an electrician to ‘fix your house wiring’, but they may only look at it, and not remove or even probe at any power sockets, but from the mere fact you’re saying something is sparking when you’re plugging it in (maybe even making a video of it), but not even making a scan of under-wall wires even if you’re giving the equipment for it, they have to fix it in one go and if that’s not good instantly, everyone you know will have a go at them.

That’s still a bad analogy, but sometimes people like thinking in analogies.

There’s no undirected ‘fix everything’ procedure. Issues have to have information provided before they can be fixed.

Oddly, lots and lots of people seem to ‘like to work with it’.

If you’re saying unstable, canary server versions having a number of issues that are yet unfixed as they’re unreproducible makes a system feel fundamentally broken, I really wonder if you ever used any canary builds of any software at all.

If there’s other concerns you have, post them with as much info as would be needed to induce them from a clean state. Don’t say ‘just join a server and it happens :man_shrugging:’, provide detailed steps. Look at some of the solved topics in #general-discussion:1s-reports and #general-discussion:bug-reports for what you should be posting.

2569 Did not fix the issues related to peds spawning inside cars.

I also have been testing the latest version today with 2784 and can confirm even with only these running.
mapmanager
spawnmanager
fivem

Once we get past 9 people Blimps and helicopters start falling from the sky and peds start despawning leaving their cars bugged.

Are you able to cause this in any place/way with less players? 2784 fixed an occurrence I had of leftover vehicles with 3 players just fine, but I can’t have ‘9 people’ on demand in the same area to test.

Even with a server full of people i have trouble getting 10 to help test. So i know its gotta be tough for you. Every time we were below 8 everything was smooth and great. After that it varied 9 10 11 when it started happening. Cars and air craft would lose their driver. Two main areas it was easy to see was near pillbox medical center (( a heli copter spawns and drops) and on the highway overpass near the vanilla unicorn. (this is where blimps come to a landing.

After about 10 people were on if you were in these two areas it was almost as if one of them would re spawn and drop as people would spawn into the server. Meaning one of the already spawned blimps would disappear then reappear in the sky and fall down again.

And the recent fixes made no difference at all in your scenario, or did it increase the minimum count to experience issues?

I cant compare due to the only testing ive done with the latest version was today on our dev server. Before i pushed to the live server i added a script to disable air traffic. I will say though running it today ground traffic has been much smoother then normal and was honestly having a convo about it with someone just now. Its what made me think to check back on this thread. But currently only 25 people on the server right now since we are a US based server. So will know better tomorrow during prime time.

Also experiencing this issue, running 2784 (also experienced on previous builds). Haven’t been able to reproduce the issue in our dev server, but that could be due to not having several people to join and be in the area. In the main server, I cannot recall it happening with less than ~10 people on. Mostly happens in peak time periods.

I have attempted to despawn the ped’s by running the following code on vehicle spawn, with no luck:

local maxIndex = GetVehicleMaxNumberOfPassengers(callback_vehicle) - 1
for i=-1,maxIndex do
	local ped = GetPedInVehicleSeat(callback_vehicle, i)
	if not IsPedAPlayer(ped) then
		print("Trying to remove ped from seat "..i)
		local model = GetEntityModel(ped)
		SetEntityAsNoLongerNeeded(ped)
		SetModelAsNoLongerNeeded(model)
	end
end

In a previous iteration of that code snippet, I also tested to see if the vehicle seat was free using IsVehicleSeatFree which always returned true, even if there was a PED in the seat due to this issue.

Is there anything I can provide that could assist with debugging?

If even you can’t reproduce it while having the issue, how are you expecting it to get fixed at all?

What OneSync settings are you even using?

I’ll ignore your sarcastic remark and put it down to you being jaded. No, I haven’t been able to reproduce it, however that does not mean it cannot, or should not, be fixed. It means that with enough data, which I stated I am willing and open to provide where possible, I would hope we could identify a reliable method to reproduce the issue.

Given that the only common thing I can find across all servers experiencing the issue is that it seems to only happen when player counter is greater than 10, I am attempting to get more than 10 players into the dev server in an attempt to reproduce the issue.

We are currently using onesync legacy. However, I have proposed we test the new “onesync on”

Right. However since I have literally no idea what can even cause this at all, since population spawning has no relation to any code written for the project whatsoever and is likely a coincidental interaction with some other logic that is being used, without any actual repro/small-scale logs this can’t be fixed.

Log sizes and correlation for ‘more than 10 players’ are extremely nasty to look at, even 3 players’ logs + a server log are already a stretch for a real-time distributed system, so without a repro case that involves 3 players or less being found this issue is simply not getting fixed, since there’s no way to get data from that many connections, even if I were able to get that many players at exactly the right positions on demand, it’s impossible to look at that many giant distinct verbose logs.

There’s probably something script-related that can be done to induce the exact same scenario, but that first requires finding out what exactly happens at all (e.g. object ID reuse? incorrect order for migration? peds not being removed correctly and having an invalid current-vehicle index?)

How about legacy + onesync_forceMigration enabled?

I have also suggested we toggle forceMigration after the issue occurs to see if that enables us to at the very least, kill the ped or drag them out of the vehicle

After disabling all npcs for some reason they still spawn, usually in a plane or helicopter. I have a script that fully disables them not sure if this is an issue with the artifacts or not

Can anyone confirm an older artifact that dosent have this npc spawning in owned veh issue? Using 2430 and having the issue

As a bit of testing that I noticed with the peds.
/dv doesn’t work for the vehicle if in driver seat
/dv from the passenger seat, it will make the ped driver get out, only for the person that did the /dv, that person can get in the vehicle and then anyone can drive it. (says client side issue to me)

After you get the vehicle back from a ped, it won’t disappear. But if you use vMenu, you can use the vehicle spawner to “replace vehicle” and it will remove the affected vehicle, and allow you to delete the new one.

Edit: My thought is that the vehicle spawner for peds is overlapping vehicles, so when you DV one of them, it leaves the original vehicle. Might have to do how the client is identifying the difference between entities that are local.

Ok im having this issue aswell like today this has happened even getting peds spawn on the roads like 5 to 6 sometimes more which makes traffic stop. U can’t beat them up only way to kill them is a gun bullet to there head but it still makes more ai peds spawn on the road and stand there not sure if anyone else has had this issue but I am