QuoteDisplay MoreOriginally posted by goodone
ppl, u must understand that freelancer is UDP software, not a TCP one. there is a distinct difference between those two, and in simple terms it can b explained as: server doesn't wait for packet acknowledgment from client, since it doesn't have to deal with such information, because of engine architecture. typical TCP is starlancer multiplay, and ppl who played it know what means to have a bad conn while on it.
therefore, fl server doesn't suffer from bad connections of certain players, even if there r many of those ppl (which r not in todays technology surrounding). yes, server could crash if there r many ppl with conn problems, but we r speaking in terms of 50 ppl getting packet losses at one time, and that doesn't happen on any fl server today, nor i think it had happened ever.
so, let's not confuse server crashing and u seeing ppl fly all around u with no sense. lag/ping problems affect only gameplay, ppl running from one spot to another, missile/torps/mines tracking and similar issues, not server stability, at least not on a greater scale such a crash is.
those server's crashes we experience happen because of a third reason, and we should realize that, not to spam this thread anymore by lag/ping issues.
anyway, it may b that sol has some architectural errors that make fl engine blackout when it experiences ridiculous messages from a certain client, like: i'm in a planet and dks getting on me is too (because some wh in sol push ships to planets before they turn on to warp; op u should check that - witnessed many times, mizi was in that situation today twice when server crashed; on 2nd crash we were the only ones online), or i'm in a trade lane still while i am docking earth. there r a few other situations that r unique to sol and could affect server stability in a great way, since server doesn't know how to respond to such crazy situations.
ofc, fl engine is not perfect. it may just b that sol is too huge for engine to comprehend all calculations for multiple players in it having dks on them, and it crashes as a result. this would b a fl engine flaw and we could not do anything about it, except getting sol a bit smaller and with fewer details (and if u ask me, that would b downgrade, not an upgrade). imo there is a smaller possibility crashes happen because of such flaw and i think problems result from architectural errors.
i have a few more theories, some involving fl engine and some not, but i will not speculate on those until i get more info. anyway, op, i hope this information will b of use to u, and i will continue reporting exact time of the server crash.
imho it would b good if others behave similarly if possible.
agreed on the most. Want to add that UDP is connectionless and that packages can be send also if one part is missing. Packets are not transmitted in the order they should there is no rule for it with UDP. And the application has to deal with it - and this at least can cause bad server integrity.
But nor you neither OP can change this. It depends on the UDP architecture.
And it could be that sol causes these problems as it is very big and maybe near the limit of the system size for FL. To move at limits doesn't do well a long time^^
