XPSP2 will limit your max. connections/sec


Recommended Posts

For those of you who apparently didn't know:

All P2P apps open and close a lot of connections concurrently, especially as they begin to hook up to their respective networks. Due to the inherently unpredictable nature of the networks many, if not most, of these connections will fail because the users have turned off the software or the IP addresses have changed.

As it works now this currently buggy beta security measure invariably interferes with legitimate P2P applications but this problem will be worked out so that both can coexist. For example, digitally signed software on a trusted root may be given a pass, but that check has yet to be built in because the throttle is still unfinished.

So: Don't panic just yet. ;]

Aaron.Walkhouse

www.bearshare.net

P.S: P2P is just as legal as a VCR and will stay that way.

I, for one, find plenty of legal things to share and download with ease.

this worked for me

http://www.lvllord.de/4226fix/4226fix-en.htm

After the "EventID 4226: TCP/IP has reached the security limit imposed on the number of concurrent TCP connect attempts" drove me almost crazy, I used the day to create a fix for this "feature" (we can discuss about).

Unfortunately there exists no REG-key which could easily be set (would be to nice, right? *smile*). The file TCPIP.SYS in the directory C:\WINDOWS\SYSTEM32\DRIVERS and C:\WINDOWS\SERVICEPACKFILES\I386 has to be changed (system dependend eventually in C:\WINDOWS\SYSTEM32\DLLCACHE, too).

Needed things:

- XP SP2 RC1/RC2/RC2-2 (Build 2096/2149/2162)

- patcher

- a small amount of time

What is been done:

To say it easy: the before 10 half-open connections are beeing increased to 16,7 mio and the CRC is been corrected. And that's it!

Comment:

The method described here, should only be used by users, who know how to handle all the described. With the download of the here published program the user know, that changes are made on third party files. For damages in every kind I cannot be hold responsible for. Indeed, tests with build 2149 and 2162 worked fine here (build 2096 I couldn't test because of missing servicepack). However, nothing is impossible.

Info: When error occurs, the patcher can change the TCPIP.SYS back to the original!

Instruction:

Copy the TCPIP.SYS from C:\WINDOWS\SYSTEM32\DRIVERS\TCPIP.SYS to a new, clean directory (e.g C:\FIX) and copy the patcher to the same. Execute the patcher. It will automatically find out the build of your TCPIP.SYS and if it's already patched and ask, if it should patch/undo it.

Now restart your computer an press F8 short after the Bios is done and start in securemode (don't know the right translation!).

Then we go into the directory C:\WINDOWS\SYSTEM32\DRIVERS and overwrite the existing TCPIP.SYS with our patched one and the repeat this with the directory C:\WINDOWS\SERVICEPACKFILES\I386 and system dependend eventually with C:\WINDOWS\SYSTEM32\DLLCACHE.

  • 2 weeks later...
theres no reason they should worry about it, the only time you have a large number of connections is on P2P , which isnt legal , therefore they arent going to trouble themselfs to let you download porn... i dont blame them,

Have you thought about any legal reasons? How about querying 20,000 game servers quickly? You need to in order to get a list of game servers in your server browser.

People don't get it. This is a good thing. It doesn't affect P2P or anything at all.

It only affects connections that will never happen because the host is unreachable (=dead) so there would be no connection in the first place anyway. It just limits the attempts per second to try and reach those dead hosts again. If anything it will make your connection faster because less bandwith is being wasted on waiting on responses and resending of connection requests from/to unreachable networks.

If you change the value you're connection might actually get slower because it gets satured with useless connection requests.

For once MS did a really good thing and all the kids are crying like if someone stole their sweets.

Edited by thop
It only affects connections that will never happen because the host is unreachable (=dead) so there would be no connection in the first place anyway. It just limits the attempts per second to try and reach those dead hosts again. If anything it will make your connection faster because less bandwith is being wasted on waiting on responses and resending of connection requests from/to unreachable networks.

i totally understand where you're coming from here but i think microsoft managed to stuff that one up aswell. do me a favour, look at your current netstat. the oh so stereotypical microsoft winsock api has exactly what we've come to expect.

out of my netstat output 2 connections were useful (msn messenger and the connection im using to load the images on this page) 8 connections were dead but not properly closed . if i understand this correctly sp2 would therefore deduce that i can use a whole 2 more connections on dead hosts.

if i had to choose between network saturation and rapid network scanning (like in non udp game server finders) i think i like the rapid network scanning :D

Is there a way other than the lvllord's patch that we can manually edit the tcpip.sys file ? (btw: the patch works with RTM 2180; at least xp boots up normally with patched tcpip.sys but could not test it with a p2p)

I mean how can i edit tcpip.sys file myself ?

Also what's the original value in SP1's tcpip.sys ? (i want to implement this value to the SP2's tcpip.sys file)

for those who want to test it out, lvllord has a batch file on that site

@echo off
for /l %%a in (1,1,20) DO start telnet 192.168.0.254

where 192.168.0.254 is an ip in your subnet that doesn't have a computer attached to it. then you can look for the EventID 4226 :D

Oh my god. I just read this whole thread, and I am absolutely sickened that people aren't even reading previous replies, and information posted right within this very thread to counter some claims made. It makes me want to cry.

I feel your pain, but reading 10 pages is just not something some folks have the patience for. Sorry.

Bottom line is that as long as you can turn stuff like this off, the world will still be ok for most of us. :)

Those manual instructions does NOT say how can i edit it manually...

I can use a Hex Editor to edit (i assume) but don't know where to start...

If you can use a hex editor (don't assume too much), that page gives you the values (& their offset position) to edit - but only specifically for versions; 2096, 2149, 2162 & 2180.

In further reading, it almost sounds like this "feature" is misunderstood. I don't think it is designed to screw you up and goof on your peer to peer, but is really just supposed to help protect your system.

Anyone else get that read from the data out there?

Thanks,

BK

This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.
  • Posts

    • Turbo Pascal was my first real programming experience more than 30 years ago at university. I mostly taught myself from the included examples and help documentation, because the university only taught the basic syntax and philosophy of Pascal, without going deeply into Turbo Pascal’s advanced features. I still remember when I discovered that I could embed assembly language directly into Pascal code, call BIOS functions, manipulate screen memory, use mouse interrupts, and control peripherals from my programs. That opened huge doors for me. Programming back then felt really fun, direct, and close to the machine. What I loved about Pascal was its readability and the almost instant compile time. Turbo Pascal was an amazing environment, but unfortunately Turbo Pascal for Windows 3 did not feel like it fully carried that legacy forward. Later, Delphi got things back on the right track after the messy transition to TP for Windows. Sadly, Delphi suffered from years of uncertainty as it moved from Borland to CodeGear and then to Embarcadero. That instability made many developers lose confidence in it, even though Delphi itself remained a powerful and productive tool. I still work with Delphi from time to time, but I definitely miss the old days of Turbo Pascal.
    • I hope this encodes in to AV1 or AV2 as currently tiktok uses h265 and h264.
    • Qualcomm reportedly in talks to build custom video chips for TikTok parent ByteDance by Karthik Mudaliar Qualcomm is reportedly in advanced discussions to provide custom chip-design services to Chinese tech giant ByteDance, the same company behind TikTok. According to a report from Reuters, Qualcomm could be involved in designing custom silicon tailored for ByteDance's massive data-center workloads. If it goes through, the deal would make ByteDance one of Qualcomm's early anchor customers for its fastly growing custom chip-design division, For years, Qualcomm was the king of making smartphone processors and modems. The company has also been moving into the PC ecosystem and other formats such as on-device AI for Android XR headsets. However, this particular deal is about Qualcomm's custom Application-Specific Integrated Circuits (ASICs). For a platform like TikTok, ByteDance needs hardware that can help it ingest, process, and serve billions of short-form videos daily. Generalised hardware is no longer the most cost-effective and efficient route, which is why ByteDance is trying to develop custom Video Processing Units (VPUs). VPUs designed specifically for ByteDance’s algorithmic needs could drastically reduce data-center power consumption and improve encoding speeds at an unprecedented scale. The underlying tech behind these processors is actually from Qualcomm's recent acquisition of AlphaWave Semi, a high-speed connectivity specialist company. By combining AlphaWave’s high-bandwidth IP with Qualcomm’s architectural expertise, the company could begin mass production by the end of 2026, if the talks go through. All this also comes at a time when U.S.-China tech relations have dwindled. Escalating trade frictions between Washington and Beijing have severely impacted the export of high-end AI chips from U.S. firms like Nvidia, AMD, and Lam Research. Yet, the Qualcomm-ByteDance discussions show that U.S. tech companies are still actively seeking growth avenues and are open to doing business with China, where regulators still permit. Reuters notes that the outcome of this deal could be uncertain, and ByteDance might also seek partners other than Qualcomm. via Reuters | Image via DepositPhotos.com
  • Recent Achievements

    • First Post
      Almohandis earned a badge
      First Post
    • Rookie
      DaviKar went up a rank
      Rookie
    • Dedicated
      HidekoYamamoto94 earned a badge
      Dedicated
    • One Month Later
      timbobit earned a badge
      One Month Later
    • One Month Later
      nates earned a badge
      One Month Later
  • Popular Contributors

    1. 1
      +primortal
      456
    2. 2
      +Edouard
      164
    3. 3
      PsYcHoKiLLa
      117
    4. 4
      Michael Scrip
      83
    5. 5
      Xenon
      69
  • Tell a friend

    Love Neowin? Tell a friend!