[How-To] How to speed up the auth login in few seconds!

[lollo_9_1]

Well, such a thing is quite easy, and I know it since 2011~2012 (when the first 30k game cores were released to be more precise).
You can find some diff patch files of that time for 30k game cores on the web that do this, but I will explain to you how to do it via source code editing.

Before all, why the 40k game core files have a login more slower than usual? That's because ymir implemented an additional encryption security using cryptopp, and at login, they added some key agreement checks, which ended up the login to be more slower.

You can actually disable such a feature, so that you can have the login fast like the old game revisions had. (r2089m or before)
Disabling it will actually remove the "key agreement" issue, and will encrypt the connection with a static xtea key instead of some dynamic mixed cryptopp algorithms.

The edit is quite simple:

Via server, you have to go and edit ^/Srcs/Server/common/service.h
commenting

#define _IMPROVED_PACKET_ENCRYPTION_

to

// #define _IMPROVED_PACKET_ENCRYPTION_

Via client, same story for ^/Srcs/Client/EterBase/ServiceDefs.h
commenting

#define _IMPROVED_PACKET_ENCRYPTION_

to

// #define _IMPROVED_PACKET_ENCRYPTION_

Additionally, in the client, you must also edit the archaic pong from ^/Srcs/Client/UserInterface/Locale.cpp like this:
find

#define LSS_SECURITY_KEY    "testtesttesttest"

change it to

#define LSS_SECURITY_KEY    "1234abcd5678efgh"

Video of a quick login: (using a second rate connection, and with 4810 programs opened simultaneously)




Here how the code should be at the end:


Enjoy your abione.

[Lauling]

this security option what we have to disable, what make that?

[TrueLev]

It's good and beautiful, but I want to know it too
I think Ymir did not make it "just because".

[kstmr]

Originally Posted by TrueLev

It's good and beautiful, but I want to know it too
I think Ymir did not make it "just because".

umh , umh , ymir did it inorder to protect the client a bit more but there was some small problem in the code , the problem is where is that bug.
you have a choice :
1- less protected yet faster and unbugged. (2/3)
2- more proteceted yet slower and bugged. (1/3)
i would go for 1 the choice is yours.
btw,this protection won't change anything unless you are dealing with someone who has a lot of free time.

[lollo_9_1]

Originally Posted by TrueLev

It's good and beautiful, but I want to know it too
I think Ymir did not make it "just because".

As written in the first post, in the past (r2089m or before), the connection between client and game was encrypted using a xtea key specified in the launcher. In the recent revs (30k, 40k), the key is specified by the server after the player logs in, and encrypts the connection using mixed cryptopp algorythms.

A man in the middle can sniff the packets either _IMPROVED_PACKET_ENCRYPTION_ is defined or not, since people can easily emulate and crack the code (since it's leaked). So you're doomed either cases.

People also created client-less bots for both circumstances, and you also can find a ymir script that does the same.

There's a much high possibility of being hacked running a fake launcher.exe received on skype, via team viewer, or sql injected from the website.

Anyway, a scenario in which you could get a man in the middle is kinda improbable. Someone should be between your server and internet. Something like an illegal serverfarm sniffing its own servers. (there are other ways, but this could be the most probable in these days)
The worst problem you could get from it would be that the sniffer can get all the players' passwords that tried to log in, and all the normal/private/guild messages.
It can't breakthrough your ftp or database via metin2 processes, just to say.

[redryan]

Still, cant see why you simply didnt removed the client timers.
Theres a timer when you choose the char, you just remove/make it shorter and done, you have the same login speed without disabling improved_packet.

Also, doesnt matter at all those programs open, metin2 uses really low traffic, so the login process is just some kbs. Even if you're downloading anything at mega (using most of your bandwitch) the speed isnt affected.

Actually your pack organization can affect more the speed then or bandwitch.

With that said, only with timers removed i got a 8s (since login to in-game char) login time in a official-like client and 6s login time with a custom one.

As you see in your video, your client hangs for about 3 seconds when you choose the char to when it actually starts loading.

Tl;dr
You dont need at all to disable it.
Remove client timer after choosing char

[lollo_9_1]

Originally Posted by redryan

1) Theres a timer when you choose the char, you just remove/make it shorter and done, you have the same login speed without disabling improved_packet.

2) Also, doesnt matter at all those programs open,

3) metin2 uses really low traffic, so the login process is just some kbs.

4) You dont need at all to disable it.
Remove client timer after choosing char

5) Actually your pack organization can affect more the speed then or bandwitch.

1) That delay is in there to execute the selected.gr2 animation to its fullest. It can be removed, so that the execution time can be shortened of few seconds even more.
It's unaesthetic for many, but I could add a how-to for that too.
2) Character/World rendering will be slower, so much that the select phase window will appear a bit delayed.
3) Faster the ping, faster the login response. You can't compare a 3ms of ping connection with a 60ms one. (which you can gain 1.5s and even more; login phase won't have any delay at all)
4) This how-to is primary for auth login, not character selection
5) People could actually map the entire client on ram disk if they have enough ram, so that they won't have the limitation speed of a hdd, and it's also way too faster than an ssd too. (mainly talking about the uiloading phase)
It uses no bandwidth in processing the client\pack files, except if you use type 3-4-5. (packing a big pack entirely with type 5 could actually affect the network)

[TrueLev]

To disable timer:
introselect.py

Serach for this: if app.GetTime() - self.startReservingTime > 3.0:
Rewrite 3.0 example 1.0

(With 1.0, you will see the full animation)

[.Liron]

Thank you for this release. It is quite useful.

Mfg .Liron

[lollo_9_1]

Originally Posted by TrueLev

To disable timer:
introselect.py

Serach for this: if app.GetTime() - self.startReservingTime > 3.0:
Rewrite 3.0 example 1.0

(With 1.0, you will see the full animation)

I tested changing

			if app.GetTime() - self.startReservingTime > 3.0:

to (just to keep a valid block in there)

			if True:

And the animation was still executed till the end, but without the 3s delay after that. So it's ok.

[Leibauer]

Thank you.

[-Exploit]

Originally Posted by lollo_9_1

I tested changing

			if app.GetTime() - self.startReservingTime > 3.0:

to (just to keep a valid block in there)

			if True:

And the animation was still executed till the end, but without the 3s delay after that. So it's ok.

Vielen dank.

[fcsk_aim]

**** tutorial , and useless.
I tested those things since 2014.

If you do this :
- Unknow packet headers error (kick players) Server side
- Sequence errors (Kick players ) Server side
- Buffer overflow errors (Kick Players) Server side

If you disable this improved_packet this code will be valid one

sys_err ("desc buffer mem_size overflow. memsize(%u) write_pos(%u) iSize(%d)", m_lpOutputBuffer->mem_size, m_lpOutputBuffer->write_point_pos, iSize);

Just disable improved_packet_encryption and install offline shop. You will get buffer overflow prolems.

I disabled this improved_ since 2014, and to fix these errors you need to reactivate seqeunce(Actually is incomplete) and also you need to make a buffer_adjust function, like me.

JUST A WARNING. **** tutorial, never do this.

[lollo_9_1]

Originally Posted by fcsk_aim

blabla

You what? Keep stealing work like you always do, but hide yourself before speaking again to me.
There's no way you tested it in 2014 if you were asking in 2015 how to compile the source files. HAHA (even if you did, nobody cares since you got bugs)

0 problems of packet mismatch. 0 kicks. All perfectly fine so far.

I have people who have the offlineshop system (purchased from ken, not stolen work like the one you have) and the _IMPROVED_PACKET_ENCRYPTION_ macro commented. Result? 0 problems of packet mismatch. 0 kicks. All perfectly fine so far.

NOW GET OUT OF HERE AND NEVER SPEAK TO ME AGAIN.

[fcsk_aim]

Originally Posted by lollo_9_1

You what? Keep stealing work like you always do, but hide yourself before speaking again to me.
There's no way you tested it in 2014 if you were asking in 2015 how to compile the source files. HAHA (even if you did, nobody cares since you got bugs)


0 problems of packet mismatch. 0 kicks. All perfectly fine so far.

I have people who have the offlineshop system (purchased from ken, not stolen work like the one you have) and the _IMPROVED_PACKET_ENCRYPTION_ macro commented. Result? 0 problems of packet mismatch. 0 kicks. All perfectly fine so far.

NOW GET OUT OF HERE AND NEVER SPEAK TO ME AGAIN.

Let's talk ajust about topic. Nobody care about our problems.
Some of my customers tested this, 3 of my customers tested with IMPROVED_PACKET disable.

And always buffers get overflow. All buffered packet's will overflow.
To fix this i used Libuv for auto adjust buffers

Just try it, when you server will reach 100 players. Buffer overflow will come. Belive me , is tested by 3 of my customers.

And about stolen work, if you will answer me on skype i can send you proofs and you can see is nothing stolen from you.