Works like a charm. I've often wondered why my Linux desktop never had the retry issue like my Win7 box does on the same network. I just chalked it up to Linux having a superior network infrastructure (esp the TCP stack), or possibly the fact that the Linux machine has a live IP (no external firewall). The days are long past when I'd run a Windows box that way.

I had no idea MS had the timestamps off by default. And that seems like something I really should have known before now...

Kudos all around!

---

This really helped all my nodes running BOINC v7.0.X.
I went from over 2000 AP tasks trying to download across 150 machines to less than 75 active transfers. I have stopped the program I was running to keep retrying the transfers and now all of the downloads are completing on their own.

I have to believe that not having so many TCP connections aborting would take some load off the download server(s). I hope someone is able to create a sticky note for this because all machines running MSFT Windows OS variants should really apply this one.

---

Has anyone tried disabling TCP fix after relocation of servers?

Has anyone tried disabling TCP fix after relocation of servers?

I have 2 Window's computers that I did not do the fix on, and they are just fine, one is XP, the other is Vista. Downloads fly, uploads cruise right through, and report just fine. No errors.

---

Has anyone tried disabling TCP fix after relocation of servers?

There's no need to do it, the "fix" just makes Windows use features, which Linux is using by default. Also it does not only help with connection to SETI servers, but all overloaded servers/connections. For me for example it also helped in case of Rosetta, where I was also getting HTTP errors every now and than, depending on the load on their servers. And it disabled backoffs on my BOINC 6.12.34 machine, however that is possible, so no, that will stay here as it is.

---

No it isn't a "fix" it is the way it is supposed to be. MS decided to "break" it, no one can say why.

---