View Full Version : FRE's Clientside.....
steweygrrr
24-02-04, 22:30
....Think again. FRE's are serverside. And to back that claim up I am going to relay an experience that happened not 2 mins ago. My clannies and I are on our way to a clan hunt on Uranus in an APC. We cross the zone line into g_07 and I crash. On returning I find that everyone else has crashed too due to an FRE. Is there something KK ain't telling us about these bugs or is it just a random coincidence that 10 people FRE all at once?
Some zone lines are bugged. Then they get fixed and other zone lines get bugged. You seem to have found a bugged one. :o
what's a FRE?
(I've got my head stuck into NLpro so much, i dont know much else.... :D)
steweygrrr
24-02-04, 22:40
Originally posted by Glok
Some zone lines are bugged. Then they get fixed and other zone lines get bugged. You seem to have found a bugged one. :o
Its happened again
we just all FRE'd again
FRE BTW is Fatal Runtime Error
after the SpongeBob/DF wedding on uranus i jumped in the back of a random reveler and went on a trek. on crossing a zone the other 2 ppl (driver and runner) both fataled. i did not.
i think its just chance in a vehicle, just zone and hope.
steweygrrr
24-02-04, 22:49
Next issue....just been kicked for a cheat violation o0
The past 3 days every other time I access a CityComm I FRE. I thought this problem was sorted. But hell we still have bugs like the wrong item display in your inventory after death bug... which has been around since... BETA. You think after almost two years shit like that would be fixed.
Psyco Groupie
24-02-04, 23:08
had the same thing in an apc yesterday ... tis weird
its not serverside tho, but its the same thing making everyone crash
I think it's problem with the APC, like when 5 or more people in it -> zone -> everyone crashes. Personally I don't have any experience of this but this is that I've heard.
Could well be some error on the server, you cant say for sure that it's not server-side. I know the crash is clientside but the server could be sending out the wrong kind of packet which then causes the client to crash - maybe this happens more often when the server is under more strain (with more players).
I wouldnt necessarily put all the extra crashes I get on Saturn down just to my client not being able to cope so well with more people online [though tbh I never play on there anymore for this amoungst other reasons].
alot of times when me and a couple friends go and zone from pp1 to p3, all 3 of us will FRE...
kinda weird when you ALL crash...
steweygrrr
25-02-04, 21:11
Maybe theres a hardware issue?
Psyco Groupie
25-02-04, 21:17
Well .. a crash cant ever really be hardware if alot of people are having it, more that something in their hardware (something common like ram size) is causing the software to trip itself up ...
Take a look at the CPU priority thread .. seems to be helping most people
Mighty Max
25-02-04, 22:11
Think again. FRE's are serverside
Thats BS tbh. No serverside action can trigger a access_violation on your machine. That can only be done by software running on that machine .... so the client.
My theory on Fre's goes as follow
-Fre's have nothing to do with the 3d engine code.
-They do infact have to do with the netcode
-Lag is directly proportional to FRE's
-German players hardly complain and reakktor employes say there are no fre's
Thus I conclude that FRE's are the result of sloppy netcode, you recieve a shitty burst packet and the netcode cant make sense of it. (Eg, divedes 1 by 0)
Thus you crash
EDIT : I bet a 4 slot CS that this is it...
QuantumDelta
25-02-04, 22:49
Originally posted by G.0.D.
My theory on Fre's goes as follow
-Fre's have nothing to do with the 3d engine code.
-They do infact have to do with the netcode
-Lag is directly proportional to FRE's
-German players hardly complain and reakktor employes say there are no fre's
Thus I conclude that FRE's are the result of sloppy netcode, you recieve a shitty burst packet and the netcode cant make sense of it. (Eg, divedes 1 by 0)
Thus you crash
EDIT : I bet a 4 slot CS that this is it... Sounds about right... thought I think maybe a couple other factors.
Since I get 25 ping, never crashed before my upgrade, upgraded, started crashing "somewhat" regularly again (pity, they don't want our callstacks atm...)
Mighty Max
25-02-04, 23:10
God, i have a simple test to provve you wrong.
Just do it if you are interested:
- Go kill anything.
- Open the corpse move your mouse ove one of the loot items.
- a tooltip will appear (part of the 3d engine tho)
- wait till the corpse disappears
What will happen?
You crash with a FRE before you see the corpse disappear. Thats because of the tooltip causing the access_violation.
Test against my theory:
- Go kill anything.
- Open the corpse move your mouse not over an loot item
- no tooltip will appear
- wait till the corpse disappears
And? No Crash.
A simple thing. I've reported this bug after #196 hit us. And just added this test to my report since i thought of a method to proove me right :p
That FRE i.e. has nothing to do with netcode, rather with the displaying/item management. Yes there are probably additional FREs in the NetCode. But FREs are WILDCARD errors.
PS: Everyone can try this simple test.
Mighty Max, that is just one bug you are reporting, which can be repeated. I guess dodgy network packets are more likely to be sent randomly. That test you have does not infact prove that G.O.D. is wrong in his assumptions.
How can you explain that I receive about 20x more fatals on Saturn than I ever do on pluto, when the only noticeable difference on Saturn is the number of players - and yes I can still get fatals zoning into a near empty zone on Saturn?
I agree totally with the theory that the fatal errors are more in the netcode than anything else. Sure other things _can_ cause them, but I believe 90% of the time it is the network at fault. (Believe me, I used to fatal 20+ times a day, and I fixed it by making certain adjustments on my end of the network)
Virtually all the fatals I ever receive are when I: 1) cross a zone border or 2) open citycom (although citycom is rare)
Neocron crashes on occasions when it _requires_ something from the server in order to continue, not when the client is trying to display items.
All the fatals I get result in an access violation inside NTDLL.DLL - which is no doubt where the network access is being performed and NOT anything to do with the graphics engine. Surely those would appear inside my directx dlls? Or a dll associated directly with the neocron engine.
Mighty Max
26-02-04, 00:57
Thats the problem. ntdll.dll is nothing but the basic runtime library for Windows NT.
All definitions of winbase.h in the windows sdk refer to this DLL. It is all common DLL every Windows program uses, and has nothing to do with the network.
That you might expierience more FRE's at Saturn, can be caused by the more players there are and more actions of players that are taken. Resulting in a greater chance of these actions hitting the nerf of NC and letting it crash in an FRE.
And yet i didnt say that there are no FREs included in the netcode, im just saying that it is not the only reason like it was stated by G0D. And infact the server cant crash your Client. the translation and running of the commands from the server does.
Originally posted by Mighty Max
And yet i didnt say that there are no FREs included in the netcode, im just saying that it is not the only reason like it was stated by G0D. And infact the server cant crash your Client. the translation and running of the commands from the server does.
And if the server makes a mistake, sends your client a load of crappy packets it doesnt understand it makes your client crash itself. That is what everyone is trying to say here about server crashing clients. Or really how the client handles erroneous packets sent from the server which is part of the netcode - and maybe the quality of the data that the server manages to send out is something that is server-side, based on the servers code/handling of multiple sends [edit: and perhaps that is based around some limitations on the actual hardware they have up there].
I'm not trying to say the FRE can only be caused by network code, just that the majority of the ones that seem to be random, and unfixable may well be to do with it.
maybe if the servers were more stable FREs wouldn't occur so much, but we may never find out with the great service level 3 gives us.
Originally posted by Mighty Max
All definitions of winbase.h in the windows sdk refer to this DLL. It is all common DLL every Windows program uses, and has nothing to do with the network.
So you're saying it is not possible to access the network functionality of windows through this component? Or that the deciphering on the incoming packets can not use any functions available inside this dll?
I only wonder because I had constant crash, crash, crash before when I had wingate 4.0 on my network [which incidently has run every single standard internet application and game i've ever thrown at it perfectly]. Every error was related to this dll. _Then_ I changed from wingate to coyote linux and suddenly I was experiencing _no more_ fatals.
I wouldnt say it has nothing to do with the network at all.
Mighty Max
26-02-04, 01:16
i dont agree you there.
It was stated here that FREs are not caused by the engine or anything BUT the NetCode. There is no other reason.
Let me quote:
-Fre's have nothing to do with the 3d engine code.
-They do infact have to do with the netcode
Thats where i reacted from. I think i have prooven that those two statements in compination are wrong.
I'm not trying to say the FRE can only be caused by network code, just that the majority of the ones that seem to be random, and unfixable may well be to do with it.
see? I'm nothing against that. I just seperate the netcode more from the clientcode. Else every action in this game is only done in the netcode (since it gets send and received before really executed)
The packet-mismash or lag or whatelse might cause an FRE in the majority. That does not mean tho that the FRE is done due to an ineffective netcode. It might be an bad eventhandling, not waiting long enought waiting for fixed data...
Or simply yes, a netcode error. But i only can guess. Both ways are possible and both ways are common to such things...
You see speculating about the origin of FRE has not much sense for us. There are too much different origins and having a way to increase the FRE doesnt mean we found the erro, it could still be a result of a result of a result, that compines with another result of an result .... and gets an FRE. Noone of us can tell but KK _with_ the logs we send in.
Only when you can reproduce the FRE by 100% you can successfully can say. That code has to be corrupt!
:edit: btw, going online with my cellphone at pinb around 1300 (9.6kbit) does get me the same amount of FREs then with DSL(768kbit) and a ping of 80.
:edit: btw, going online with my cellphone at pinb around 1300 (9.6kbit) does get me the same amount of FREs then with DSL(768kbit) and a ping of 80. so you are playing neocron on your cell phone now? :rolleyes:
Of course didnt mean every crash is net code. When I think netcode I also think of the way the game handles the info recieved in the netcode, hence my 6/0 example. Sorry for any confuzled-ness
I myself am a programer, FRE's just arnt a major issue like they are for neocron in any of my games. Once and awhile I do some sloppy code and am able to pin point the problem and fix it. I suspect the majority of fre's all come from a single mistake in the code; Be it net or client.
Mighty Max
26-02-04, 01:56
so you are playing neocron on your cell phone now?
I cant imagine neocron would run with less then a 56k o.O
Yes i did, and i had a good reason. I cant have DSL at every plase i get based or commanded to.... (you know there is a job i have to do (i dont play the whole day)
I of course didnt mean every crash is net code
Sorry if i misinterpreted this, but i think that did some more.
I said 'hi' to a guy once and it FREed. I said 'OMG LAG TODAY!' and it FREed.
Thats i.e. a Cleint Side prob. Really it is. If the network would fail you would crash about 20-40 secs later. It's rather that the lag added too many lines to the chat at ones/simultaneous then causing two lines using the same pointer to add themself to. One of them will fail and therefor crash. Yes that only 99% happens in lag, but is an timing/threading issue.
Thos the other example might be netcode. But i dont know if it might reproduce, or why the netcode is influented by rares, since it is only an normal item for the code
I don't care if it is net code, engine code, or any other code. Just fix it!
Why is that such a monumental thing to do? I'll tell you why:
It can't be fixed when a foundation is laid based on corrupt code.
And there you have it. It is unfixable without resorting to destroying the game and starting over.
Solution: " Let em put up with the FRE's, it's not that bad "
Mighty Max
26-02-04, 03:31
Thats ignorant Smacktard.
If you fail to drive your bycicle and fall to the ground what is the reason? Do you call the manufacture and tell them "Fix it!" on every of the following reasons?
Maybe its really the manufactureres fault ...
Maybe the ground u drive over isnt good...
Maybe you just carried to much to drive stable ...
Maybe you were drunk...
Maybe someone hit you so you fall ....
Maybe you cant drive at all...
Whatever it is, you have hit the ground and feeled the pain.
If Neocron feels the pain. It pops up an FRE. You cant tell only by the pain what caused it. It is only the effect.
Maybe it was a Neocron Bug causing FRE whenever this situation appears to every install of neocron...
Maybe it was a paging exception within Windows...
Maybe it was some driver bringing chaos in the windows threading....
Maybe it was temperature flux ....
Maybe your PC dislikes you ... :p
Maybe some hardware has an hidden defect ...
Maybe some server modifying the data-stream...
PS: Windows is now 19 years old. And i still get BlueScreens or Reboots from time to time. some PCs dont some more often. some commonly... And do i expect them to solve them all? No, because i know it is impossible, they can only do their best, and this is what i see at KK. Always when they got hold of the reason why a FRE occures they fixed it.
example for this is Gaya Mine. Some pathes ago you couldnt leave it in 60% without FRE. Now it works.
Maybe I'm a bit naive, but I think some individuals of the communiy could help greatly KK with these bug problems. It's not the first time a great company looks outside to get some help in this matter, personally I have seen an excelent work from certain persons not long time ago.
If anyone here played BG-2 and know a bit of the community at IEEAIS (now gone), specially a person called Kevin Dorner, they would know the ammount of awesome, amazing good work (bug-fixing wise among others) some persons can do for the games they like.
It's all I can say, sadly I'm no expert in these issues.
Anyway on topic; FRE's are a serious problem that KK is delaying for too long already, even if they are waiting for the incomming DoY (who isnt anyway) there are many bugs in this game that really worry me, serious worries about the future of the game.
I wont get tired of saying it, this game has an amazing potential, and it is getting ruined by bugs and technical problems.
If neocron FRE's because of a windows incompatablity... Isnt it Neocrons job to find a fix for it? I asure you windows isnt going to write a service pack so it runs correctly with neocron!
Mighty Max
28-02-04, 00:00
:rolleyes:
I wonder why the service packs fixes some of the problems...maybe that more then just NC stumbles about those errors?
I dont aspect MS writing only for NC, thats pathatic and stupid to assume that I think this way.
Isnt it Neocrons job to find a fix for it?
What do you think what they are doing? Drink a beer and lol about that FRE?
And to find an error that is there because of the windows api is decribed wrong by the MS resources is something that takes you work to find. Dont tell me that is a rare situation. Its not.
And again, i dont say its MS fault. I just want to tell here that iit is
TOO COMPLEX
for us to make speculations.
Mighty max I like the content in neocron alot, and I enjoy the community. Those are the reasons I and many others play.
The FRE problem that has been around since the beta stages is NOT acceptable in a retail game. We pay 10 dollars a mounth for KK to find out what the problem is; Neocron by far is and remains the most unstable retail game I have ever played.
the problem ISNT some one elses, Its neocrons.
Dont blame it on drives for windows, Dont blame it on sadistic PC's beucase I have yet to see a piece of retail software with the degree of stability issues neocron has.
Now with that said, you seem to be taking this as if im bashing Neocron. Im not, Its just a part of neocron at this point and if fixed would greatly increase the enjoyment most get from this great game.
"TOO COMPLEX"? Martin has a rolex, we bought him that. He better get the FRE problem under control :rolleyes:
Mighty Max
28-02-04, 09:49
If it is not the environment and only the game, why do I crash i.e. only on rare occasions?
If the codebase is so bad like you all like to say, i should crash in the same rate as you. Yet i had often a lot slower and laggier net connection....
Martin has a rolex, we bought him that
lol. Honestly, i dont care what Joerg does with HIS money. And yet i see not why this would remove the complex.
steweygrrr
28-02-04, 13:05
What I think he was saying Max is that not absolutely EVERYONE who plays Neocron has defective hardware, chaotic drivers, sadistic PC's (O_o) etc etc etc but everyone who plays NC does fatal.
I understand that this is a complex issue but I dont think it is a compatibility one. BTW my reference to hardware was concerning the servers not us because not matter your hardware setup you WILL FRE
Originally posted by G.0.D.
Martin has a rolex,
what?
No i dont O_o
:p
anyways.
SO, why doe sthe fix i posted a few days ago seem to help the amount of FRE's?
Im not technically minded that in way in the fact I dont know why its working for me, all I know is that it does seem to help. What effect does upping the process priority of NC have over Netcode or Engine running or whatever?
(In case no one has seen it --> My Stability Fix (http://neocron.jafc.de/showthread.php?s=&threadid=91752) )
Originally posted by Elric
SO, why doe sthe fix i posted a few days ago seem to help the amount of FRE's?
Your fix is interesting, but are you saying since you ran it like that you havent had a single fatal runtime error? In that case sure, we can not blame the network code at all for the problem.
I get the feeling (this is not backed up with any fact.. just some thoughts) - that neocron servers sending larger than normal packets along the internet, so they are likely to be highly fragmented by the time they get to your box (which is dependent of which route they take to your box - not your overall lag or pingtime). It may be that there is some packetloss along the way, so these packets can not be reassembled correctly - which may not be such a problem with smaller packets but when your client is expecting some huge amount of information (such as when you zone), it may crash when it does not receive it.
The only reason I'm saying this is because when I had wingate 4 (on win98) forwarding the packets up to my pc, it fatalled much more than when I had a version of linux doing this for me. It could have been that my win98 box was fragmenting the packets a great deal before sending them up and the newer linux box was not. I cant really think of what else it could be.
Maybe the increase in CPU priority gives neocron more of a chance to handle these lost packets + request new ones before it crashes?
Btw, there were many 100% cpu internet games before neocron came out + I dont remember ever having to increase priority inside quake (in s/w mode) in order to stop it from crashing/disconnecting me.
Mighty Max
28-02-04, 14:42
that neocron servers sending larger than normal packets along the internet,
To set up facts here. The complete datapackage Neocron sends is not longer then 235Bytes yet. Neocron tends to send some of them simultaneous. Givin up packets upto 1kB. yet this is under the TCP/IP packet limit.
Originally posted by Mighty Max
To set up facts here. The complete datapackage Neocron sends is not longer then 235Bytes yet. Neocron tends to send some of them simultaneous. Givin up packets upto 1kB. yet this is under the TCP/IP packet limit.
Where did you acquire this information?
Mighty Max
28-02-04, 14:50
With my work on this http://neocron.jafc.de/showthread.php?s=&threadid=74671&goto=lastpost tool.
Since it works, my assumes must be right ...
steweygrrr
28-02-04, 15:21
Originally posted by Elric
(In case no one has seen it --> My Stability Fix (http://neocron.jafc.de/showthread.php?s=&threadid=91752) )
Seen it, used it, no effect sorry
yep, for some it works, for some it dont.
I was running NC for about 2 weeks before trying this.
I have made NO changes (AT ALL) to my system in the last 4 weeks.
2 weeks before trying. Regular fre's. Maybe once or twice every 45 mins - 1 hour.
Since changing NC everytime i start it to High Priority. 3 Fatals on Pluto, 4 On saturn. Total, through 2 weeks of playing. When I forget to change it to High Priority, its almost gauranteed to crash me within 1.5 to 2 hours, if not sooner.
like i said, Im not technically minded or anything when it comes to that sort of thing so i dont know WHY it works for me and not for others, But I do know that it works.
Powered by vBulletin® Version 4.2.3 Copyright © 2024 vBulletin Solutions, Inc. All rights reserved.