Hi all,

2 x XPSP2 machines here running Boinc/Seti. No problems whatsoever.

2 x 98SE machines running Boinc/Seti and problems.

The problems do NOT - repeat NOT - surface with any machine until server contact is needed. When the XP machines contact the server, no problems. When the 98SE machines attempt to contact the server, all slows down until you cannot do a thing other than a reboot (sometimes possible) or a reset button job.

Why some people have corruptions to their file systems thereafter is NOT attributable to Boinc/Seti but attributable, rather, to the old truth - you didnt shut down your system appropriately and thus CLUNK, no data. That can happen to any system, XP included. I have had to reboot by reset button, the 98SE machines a number of times.

Thus, you can see the problem isnt actually with Boinc/Seti in total, only the routine that contacts the server at Boinc and then only with 98-ME machines, not XP machines (and I assume W2K machines are similarly unaffected).

Suggestions:

1) BOINC PEOPLE - FIX IT OR LOSE ANYONE NOT ON W2K/XP!!

2) If you do complete "C" drive images, make sure they are done regularly until Boinc people fix the problem. If you dont, Symantec took over Powerquest who's "Drive Image 7" program was marvellous. Symantec have a habit of stuffing up programs they take over (just look at NIS, formerly Atguard, a brilliant program till Symantec made Swiss Cheese out of it) and I am HOPING this didnt happen to Drive Image. If it didnt, it is the program to buy in order to do images. If not, hold onto your seats and pray.

3) If you dont want to or dont understand imaging, then dont allow Boinc/Seti to auto start, dont use it and await the Boinc people fixing the problem THEY created.

Greg.

I am investigating the issue.

I'll keep people posted on what I figure out.

----- Rom
BOINC Development Team, U.C. Berkeley

[url=http://mirror.sit.wisc.edu/mirrors/apple/ausmac/Quicktime-Movies/power-macintosh/apple-win95-crowd-control.mov]Get a MAC !

> I am investigating the issue.

> I'll keep people posted on what I figure out.

Feel free to send any tests you want to make this way. You know how to find my email address and I am on ADSL. The 98SE machines, here, are connected via simple ICS on my home lan and I am not working today so can run a test quickly. Attach to email if you like. I keep images on all my machines so killing one isnt a serious problem.

Greg.

Rom, is this the same problem that Darrel reported in alpha? Never heard about anything being fixed before release.

---

Account frozen...

Darrel reported a stack overflow issue. I'm still working on that issue as well.

----- Rom
BOINC Development Team, U.C. Berkeley

> Darrel reported a stack overflow issue. I'm still working on that issue as
> well.

Wouldnt the simple answer be a "cut and paste" of any server routines from the pre 4.05 version of Boinc to fix the problem or did something change at the server end?

See, my view of this without being able to see the code is that you have a simple:

"Is the server able to be contacted?" question expecting an immediate response. Eg, you didnt allow for the time it takes for server contact to be made from various parts of the world over the net. Then, you either dont have an answer so it sticks where it is or you have it return to ask that question all over again and not wait for the response, thus the "slowing down" effect to an eventual halt.

Should be relatively simple to track down if I am correct but I understand I well may not be.

Greg.

> Darrel reported a stack overflow issue. I'm still working on that issue as
> well.
>
> ----- Rom


I have the time and the motivation for just one help.

Funktion: NtWaitForMultipleObjects
77882873 b8e9000000 mov eax,0xe9
77882878 8d542404 lea edx,[esp+0x4] ss:09ec9e33=????????
7788287c cd2e int 2e
7788287e c21400 ret 0x14

Actually upon further review, it seems Darrel did report these types of issues as well. I thought it was related to another project that was comming online in the not to distant future and missed some of the details.

Looks like I owe Darrel and all the people who use Win9x an apology for this.

----- Rom
BOINC Development Team, U.C. Berkeley

> I have the time and the motivation for just one help.
>
> Funktion: NtWaitForMultipleObjects
> 77882873 b8e9000000 mov eax,0xe9
> 77882878 8d542404 lea edx,[esp+0x4]
> ss:09ec9e33=????????
> 7788287c cd2e int 2e
> 7788287e c21400 ret 0x14

Are you using a debugger to generate this? Do you happen to have the PDB on the machine? Any chance you can happen to get the callstack for that function call? Does this happen after the first reply from the scheduler and it is downloading workunits in the background?

----- Rom
BOINC Development Team, U.C. Berkeley

> 2) If you do complete "C" drive images, make sure they are done regularly
> until Boinc people fix the problem. If you dont, Symantec took over Powerquest
> who's "Drive Image 7" program was marvellous. Symantec have a habit of
> stuffing up programs they take over (just look at NIS, formerly Atguard, a
> brilliant program till Symantec made Swiss Cheese out of it) and I am HOPING
> this didnt happen to Drive Image. If it didnt, it is the program to buy in
> order to do images. If not, hold onto your seats and pray.

DI 2002 (or DI 7 - same thing) are good programs. I've heard the new versions have problems but you can still get new copies of 2002 ( or 7 ) on EBay easily enough.

>
> DI 2002 (or DI 7 - same thing) are good programs. I've heard the new
> versions have problems but you can still get new copies of 2002 ( or 7 ) on
> EBay easily enough.
>
>

Actually they arent the same thing. DI2002 is for 98 up to ME and 7 is for W2K and XP.

There are no problems with DI7.01 (eg you need to upgrade from 7.00).

I do not know about any versions of those programs since Symantec bought them out, however but if Symantec have followed normal protocol.....sigh....

Greg.