Olorin

Member
 View profile  See their activity

  • Content count

    73

  • Joined

  • Last visited

Posts posted by Olorin

  1. Posted

    And what the problem???

    Only exact share size&IP is very stong identities. But there are autoreplier... tag... mac!!! And the mac is the strong identity in one session. Don't you think so? Macs are sent with each tcp frame. Tag is in the users list, IP is got with standart protocol methods... WHAT ELSE??? do you need to identify a user??? That is a very strong identification method. Don't you think we are banning users only by IP... that is a coplicated method using mac, ip, nick, exact share, tag, autoreplier and route trace to a host. We a counting an ID like torrent id for file, hashing parts of a user identity to prevent a user database rob.

  2. Posted

    I have 7 hubs opened at a time. All in a LAN. Many people uses all hubs at one time, so if one of them tries to load a file, he links to me from 3-5 hubs at a time, flooding my slots. There is limiter by ID needed. ID is cummulative property of a user=[ip, all tag fields, exact share, nick]. So if one parameter like nick is not the same, the user still defined as known and linked user to prevent new connections from him. And parameters showld be rated. For example, same share or IP is a miracle. Also I can flood the same way other users and the limiter is needed even here.

  6. Posted

    I have made some limitations to solve the problem. But the problem is still actual.

    2 files at a time

    8 downloads

    no new if speed is over 1024 kBps

    It's flooding. This is not a good thing if the server has 1Gbps uplink. It likes I can flood a great LAN segment.

    Did developers read these posts??? Any comments?

  8. Posted

    So which is it? You connection cannot be 5 different things at once.

    My own uplink is 1Gbit/s twisted pare. And I use a 100Mbit/s connection to my own switch at my flat. There is no any troubles with uplink width I think. And there are no any losses in transmission. My switch is made by Asus so there are no problems even here. And my uplink is to a great metronet http://www.netbynet.ru/about_netbynet/net_map_top/ so no problems with tcp transmission.

    I have a real static LAN IP and I have another real static WAN IP so no any conflicts. All tcp/udp ports are permitted, all routes are sat, and I have "0.0.0.0 if", no wan and "manual forvarding ports" connection at DC properties.

    But the client loses the connections with the hubs if I have 20-50 sources for my download. If I have 2-3 downloads with 1-10 sources for each, Apex also loses connections to the hubs. It looks like Apex flooding itself, like too much reiterating connections are made again and again overfilling connection tables at the equipment.

    Does It connect to all souces again and again? Does Apex close the unused connections? I can't understand the nature of the problem.

  10. Posted

    Sorry for asking, but as you explain it, are you sure this you are not using some Apex mod?

    :B

    That was a great joke. :( I'm sure. As sure as I can. For example MOD has a ring icon but Apex has a drop icon. Are they?

  11. Posted

    Sorry if duplicated.

    1b4, 1b5 makes the hubs to disconnect me on downloads or disconnects itself. It likes as Apex floods the hubs or Apex doesn't keep connections to the hubs. Also it looks as Apex tries to connect to ALL of the got sources (not to first ten of them) every programmed time through all process of download. I can't really understand the nature of this process. I have up to 200 sources at a time.

    Also "remove all users from the queue", "readd all users to the queue" buttons and "readd users to the queue" list needed. Lists of sources and users it the queue to remove or readd should contain Username (IP, last speed, tag) because if I have 180 sources from all of the world I can't choose the needed ones. If I remove the sources the Apex will find them at the next search and will readd them itself - not a good idea for me. If I manual search alternatives I need a "add user to the queue" button because the automation at this process makes me delete all users again and again one(!) by one at now (1b5).

    Thank you for your time and sorry for the English.

  16. Posted

    Apex does not support kademlia. Many hub owners consider it a no no.

    Also, many ops also think multi-source is bad. But not if correctly implemented.

    I'm operator.

    I don't think ops and owners at our LAN care about multisource. The most problem is connectivity - users don't know anything about routes or connection settings at cliens. The second problem is "bad words", porno and "medved's sleng" at chats. Also ISP grows intensive and extensive ways, buying small competitors. Metro is growing about 25000wks per quarter. And we have all the problems caused. So the routing and the peering problems are more interesting... Our main links are 10Gbits ( %-/ ). The userbeasts population at our LAN comes to 100 000 workstations now - netbynet Moscow. And more than 50% are lamers! More than 95% are newbies. That is the real situation at LAN ISP market in Moscow. We have 4 large ISPs with metronets.

    And multisource is important for users, not for operators, ISPs or anybody else.

    DC users number is growing and now is coming up to 4000-10000 sorry for approximation. - Too much hubs.

  17. Posted

    DC was never supposed to be like torrent and such that allow u to download off 100 users at once.

    In fact, this would kill DC, it would mean a user with this capability cud take a slot from every user in hub.

    DC hubs vary so much in user size and they also dont have to connect to many hubs.

    Problem is a user in a few hubs, dling 5 files with 50 segements in each is using 250 users! in a small hub this would be the equiv of a slot from every user....

    you are right. But where is the problem? I can't understand the idea. Why the multisource will kill the dc? If the kademlia.dll is, if unlimited dayDC (for example) is.

    I have already found the forum discussions and even now I can't understand, where the problem?

    May be if downloading is slow the shareing time for leechers is greater? Don't forget that Apex can upload unfinished files now - this is a step to real multisource.

    after 2 hours of reading I think 5-10 links is good enough. And after 2 years of DC clients using I think "autodisconnect" should be on by default with rules rated with percents.

  18. Posted

    Definitely NOT and it doesn't have anything with hardware capabilities. It had been discussed a lot in the past so make some effort and search.

    (and there's no "get Your hands away from max segments" option in the pool..)

    sorry you are right.

    Help me searching, please. - My native is russian.

  19. Posted

    I have gigabit adapter linked at gigabit. I have 50-100 sources at the hub. But I can use only 10.

    1 source makes 1-10 megabytes per second

    10 sources make 10-100 megabytes per second.

    So real speed is 2-5 MB/s.

    And I can utilisate up to 16.5 MB/s on my 4xIDE133 striped HDD on PCI33 controller.The limitation because the 32 bit PCI33 link speed equal 133 Mbit/s

    And you with SATA300 HDD theoretically can utilisate up to 33MB/s NOW! (Real speed is not so big of course :P )

    Next generation SATA makes links with 600Mbit/s so only 2 striped HDD will be able to utilisate 150MB/s (oh... 120-125MB/s - it's critical to copper Twisted Pair). Yes, up to 150 MB/s because the current raw HDD R/W speed is 960-1100 Mbit/s - read you HDD description, read specs at manufature sites.

    Referring to NAND flash researches access time can be 50ms. I means the speed will be great.

    So the number of sources should be increased.

  20. Posted

    I have been sleeping when the master thought that the users should be at another hub and redirected us.

    Apex 1b4

    Then the Apex needed the password. So it showed the popup pass window. One window for each reconnect. So bugs:

    1) Apex doesn't check the destination hub pass at favorites (cross check hub.lan <-> 1.1.1.12 or 1.1.1.12:411 <-> hub.lan)

    2) Apex doesn't check if old popup pass windows for destination hub was created. So it should destroy old popup or use old popup for password request.

    3) Apex should not reconnect to redirection destination hub again before it get the password from user through the popup or from the favorites.

  21. Posted 

    Code: c0000005 (Access violation)

Version: 1.0.0B4 (2007-08-07)

Major: 5

Minor: 1

Build: 2600

SP: 2

Type: 1

Time: 2007-08-13 22:25:53

TTH: 7CNQCQXDGEV7S2IDTUSYEW73M722LLS7BVOOXDI


RICHED20!0x74E4C23D: IID_IRichEditOle

RICHED20!0x74E4D865: IID_IRichEditOle

RICHED20!0x74E5DE66: RichEditANSIWndProc

RICHED20!0x74E5E800: RichEditANSIWndProc

RICHED20!0x74E5EC24: RichEditANSIWndProc

RICHED20!0x74E5F1AA: RichEditANSIWndProc

RICHED20!0x74E5C331: RichEditANSIWndProc

RICHED20!0x74E42095: IID_IRichEditOleCallback

RICHED20!0x74E42C88: IID_IRichEditOleCallback

RICHED20!0x74E3EED1: IID_IRichEditOleCallback

RICHED20!0x74E7B2DA: REExtendedRegisterClass

RICHED20!0x74E4FF17: IID_ITextHost

RICHED20!0x74E51214: RichEditANSIWndProc

USER32!0x7E418724: GetDC

USER32!0x7E418806: GetDC

USER32!0x7E41B4B0: DefWindowProcW

USER32!0x7E41FD09: UserClientDllInitialize

ntdll!0x7C90EAE3: KiUserCallbackDispatcher

USER32!0x7E4201D7: CreateWindowExW

USER32!0x7E41FF63: CreateWindowExW

d:\program files\microsoft visual studio 8\vc\atlmfc\include\atlwin.h(734): ATL::CWindow::Create

d:\cvs\apexdc++\includes\wtl\atlctrls.h(7249): WTL::CRichEditCtrlT<ATL::CWindow>::Create

d:\cvs\apexdc++\trunk\windows\hubframe.cpp(67): HubFrame::OnCreate

d:\cvs\apexdc++\trunk\windows\hubframe.h(71): HubFrame::ProcessWindowMessage

d:\program files\microsoft visual studio 8\vc\atlmfc\include\atlwin.h(3078): ATL::CWindowImplBaseT<WTL::CMDIWindow=0x00DEF008,ATL::CWinTraits<1456406528=0x00000000,64> >::WindowProc

USER32!0x7E418724: GetDC

USER32!0x7E418806: GetDC

USER32!0x7E41B4B0: DefWindowProcW

USER32!0x7E41FD09: UserClientDllInitialize

ntdll!0x7C90EAE3: KiUserCallbackDispatcher

USER32!0x7E4201D7: CreateWindowExW

USER32!0x7E41FF63: CreateWindowExW

d:\cvs\apexdc++\includes\wtl\atlframe.h(443): WTL::CFrameWindowImplBase<WTL::CMDIWindow=0x000B025A,ATL::CWinTraits<1456406528=0x0006E960,64> >::Create

d:\cvs\apexdc++\trunk\windows\flattabctrl.h(882): MDITabChildWindowImpl<HubFrame=0x000B025A,255=0x0006E960,129=0x0006E748,243=0x00000081,WTL::CMDIWin

ow=0xFFFFFFFE,ATL::CWinTraits<1456406528=0x7C911B09,64> >::Create

d:\cvs\apexdc++\includes\wtl\atlframe.h(1759): WTL::CMDIChildWindowImpl<HubFrame=0x0006E960,WTL::CMDIWindow=0x0000009F,ATL::CWinTraits<1456406528=0xFFFFFFFE,64> >::CreateEx

d:\cvs\apexdc++\trunk\windows\hubframe.cpp(221): HubFrame::openWindow

d:\cvs\apexdc++\trunk\windows\mainfrm.cpp(1263): MainFrame::autoConnect

d:\cvs\apexdc++\trunk\windows\mainfrm.cpp(925): MainFrame::onSpeaker

d:\cvs\apexdc++\trunk\windows\mainfrm.h(112): MainFrame::ProcessWindowMessage

d:\cvs\apexdc++\includes\wtl\atlframe.h(1519): WTL::CMDIFrameWindowImpl<MainFrame=0x00000000,WTL::CMDIWindow=0x000081F4,ATL::CWinTraits<114229248=0x00000003,262400> >::MDIFrameWindowProc

USER32!0x7E418724: GetDC

USER32!0x7E418806: GetDC

USER32!0x7E41C623: IsWindowUnicode

USER32!0x7E41C649: CallWindowProcW

d:\program files\microsoft visual studio 8\vc\atlmfc\include\atlwin.h(3969): ATL::CContainedWindowT<ATL::CWindow=0x0006F324,ATL::CWinTraits<1442840576=0x000081F4,0> >::WindowProc

USER32!0x7E418724: GetDC

USER32!0x7E418806: GetDC

USER32!0x7E4189BD: GetWindowLongW

USER32!0x7E418A00: DispatchMessageW

d:\cvs\apexdc++\includes\wtl\atlapp.h(992): WTL::CMessageLoop::Run

d:\cvs\apexdc++\trunk\windows\main.cpp(431): Run

0x001C0118: ?

ApexDC!0x0044E0DA: MainFrame::FileListQueue::`scalar deleting destructor'

ApexDC!0x0046A7F1: [thunk]:MainFrame::`vector deleting destructor'

kernel32!0x7C83184D: GetComputerNameW

    I don't know why.

    DCPlusPlus.xml

  22. Posted

    "crypted"? i know "get", "post", "put", but "crypted"...? unless you meant "post" through a secure connection?

    as for the vote, i don't understand it (and can't find the meaning of the "else" either...).

    I can't speak english clearly because of no practice.

    Yes, ssl.

  23. Posted

    I think webserver is a great thing.

    I can search (if... if... if... too many if...) I can view logs (very-very short loggies) I can view queues if the Apex want to show them at the moment.

    Also I can even reboot the machine.

    1) use "magic quotas" as at MySQL. Quote all search requests to prevent situations when the query is at national symbols and the brouser makes a big poo with the "GET" query at address string.

    2) Use " " to "%20" translation for all national queries

    3) Make the user possible to set the code page for webserver (english are shown clearly but national symbols are not)

    4) At the future you showld make the Apex use "post" and "crypted" method for getting queries from the webinterface.