Please note my other post about BOINC 4.05 and running TWO projects on Win98SE.

BOINC fails to respond, makes the system sluggish....and needs to be stopped entirely to return the system to normal operation.

The details are in my post.

> Please note my other post about BOINC 4.05 and running TWO projects on
> Win98SE.
>
> BOINC fails to respond, makes the system sluggish....and needs to be stopped
> entirely to return the system to normal operation.
>
> The details are in my post.
>

From my other thread:

The Win32 4.05 gui client - with two projects active (67% SETI and 33% CPDN) - goes into La-la land....and does not repond to more than the first page selection in the GUI.

The system itself becomes sluggish and slow to respond to mouse input or keyboard input.

C+A+D display shows it is "Not Responding". The task can be ended and the system returns to normal.

If I choose the "Work" page as my first choice, I see a list of SETI units with expected estimated times and ONE cpn unit with an estimated time of over 672 DAYS....ending in late 2005 (which is wrong anyway for 672 days).

Even with ONE project, the v4.05 client had failed withg error messages in the past 24 hours. I did not see the errors as my daughter had cleared them. But at least it appeared to run normally otherwise.

System: Win98SE with latest patches. Intel PIII-866 / 256MB. AVG free anti-virus running - but stopping it makes no difference. Network connection is rt2400-based wireless 802.11b adapter. Connection quality 86% (Excellent).

Rom,

I urge you to be much more explicit in the news page. Something like "There is an issue with Win98 and WinME machines that can lead to DISK CORRUPTION. Do not run BOINC on these machines until further notice"

I am aware of other posts, eg:
>It could be that, as you seem to be getting at, the actual FAT corruption was >caused by the hard reset. A few others have reported some data file corruption >which could just be open files at the point of reset, but I haven't seen >anyone reporting FAT corruption. (I thought I did, but that person clarified >that his problem was datafiles, not FAT.)

The point I want to make is that in this environment surely the professional approach is to warn the unsuspecting public in case this is more severe than it sounds from the current posts? Also, the only people you can hear from are those with either multiple computers or the knowledge of how to reinstall from a corrupted FAT. Everyone else won't be able to complain since they can't use their computers!

I have a spare machine I can load Win98SE on and it is expendable if you need another guinea pig for testing.

Good luck with the fire fighting!


</img>

>I guess I am the luck one. Ive had seti running on my win98se machine since ive downloaded it. It doesnt appear to be slowing my machine or crashing it either.

system specs
Win 98 SE (installed late 1999!) with all current updates and patches
AMD K7 at 1.8ghz (not overclocked)
Norton Internet Securities 2003 running
512MB pc2100 ddr ram
40gb hd (western digital)
gf4 video
net connection tx100 net card on cable modem with 1024mb up and down bandwidth
running seti 4.05 and origional seti (neither set as screen saver as i just shut monitor off for that)

networked to 3 other machine (all running win2k sp4) and all with dedicated individual network ip's and bandwith

surprising enough, I have had it lock up on the win2k machines when im downloading and installing but not on my 98se machine

only connecting to one project.

just a suggestion but have u looked at the cooling systems inside your case lately?
the last time i seen an IPF error in kernel was usually the result of a cooling fan dying

> urge you to be much more explicit in the news page. Something like "There
> is an issue with Win98 and WinME machines that can lead to DISK CORRUPTION. > Do not run BOINC on these machines until further notice"

The problem with saying it could lead to disk corruption is that it is, in effect, rubbish. It is because the machine needs a reset rather than allowing it to shut down. That is ALWAYS at the risk of the user. The corruption isnt caused BY Boinc/Seti but as a result of that prog causing the machine to slow down. IOW, any other process could cause the same result. Thus, the user should ALWAYS be prepared for disk corruption no matter what caused it.

Dont blame BOINC if your C drive stuffs it. Hire someone to kick you in the arse for not doing proper images to ward against total data loss in the first place!

Greg.

ok so always the same problem, i made a clean install of boinc yesterday evening and let it run all the night....it done 2 wu's and stop responding, no way to unfreeze it, ctrl+alt+del is the only way.
i stop using boinc on this machine until someone find a solution.

> I urge you to be much more explicit in the news page. Something like "There
> is an issue with Win98 and WinME machines that can lead to DISK CORRUPTION.
> Do not run BOINC on these machines until further notice"

I've updated the news item. Thanks for the suggestion.

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

I wan't going to mention this to Rom, as he's fighting hard on the win95/98 issues, but I've seen this same thing on win2k with 2 projects. I ran Cpreditor for a few days with out a problem... Attached seti, and after the second context switch (preempt), it seems the machine then into the twilight zone. The video was AFU's, and until I killed boinc, my status bar was showing up at the top for example, and things were not repainting, windows were in the wrong spot, etc. This happened twice...

I detached from Cpred, and seti has been running cool for hours (after a reboot with the original problem). An interesting thing, is that after the reboot, and prior to detaching from CP, I was getting a "max quota execeded" message from the scheuler at CP - in other words, no work from cpredictor, so I was back to a single project)

The event log does show a few dialogs that VM was low (5-10), but I never saw them

> Sure.

Walt, something is going on with the data servers at SSL, so I'm going to have to send this to you via email or via http by way of my home machine.

Could you send me an email?

rwalton@ssl.berkeley.edu

Thanks in advance.

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

> Sure.

Walt, something is going on with the data servers at SSL, so I'm going to have to send this to you via email or via http by way of my home machine.

Could you send me an email?

rwalton@ssl.berkeley.edu

Thanks in advance.

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

> > Sure.
>
> Walt, something is going on with the data servers at SSL, so I'm going to have
> to send this to you via email or via http by way of my home machine.
>
> Could you send me an email?
>
> rwalton@ssl.berkeley.edu
>
> Thanks in advance.
>
> ----- Rom
> BOINC Development Team, U.C. Berkeley
>

Its on the way.

Are you sending the symbols along with the program?

I installed the Boinc 4.05 software this morning on
a machine running Windows 98 original edition, no
updates.

System specs

AMD K6-2 CPU, 333 MHz
512 MB hard disk
64 MB of RAM

No problems yet.

The machine had been running Boinc 3.19 for about
one month with no noticeable problems.

> Its on the way.
>
> Are you sending the symbols along with the program?

Yes...

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

> I installed the Boinc 4.05 software this morning on
> a machine running Windows 98 original edition, no
> updates.
>
> System specs
>
> AMD K6-2 CPU, 333 MHz
> 512 MB hard disk
> 64 MB of RAM
>
> No problems yet.
>
> The machine had been running Boinc 3.19 for about
> one month with no noticeable problems.
>
>

Isn't the new winsock required for boinc on 98? "no updates" might not be a good test (wsock2).. I forget if this was win95 or win98....

> Isn't the new winsock required for boinc on 98? "no updates" might not be a
> good test (wsock2).. I forget if this was win95 or win98....

Ug, you almost gave me a heart attack here... So BOINC_GUI.EXE links against WSOCK32.DLL, I think we'll be fine since it'll be up to WSOCK32.DLL which winsock provider to talk too.

Although I could be completely wrong here.

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

> > Isn't the new winsock required for boinc on 98? "no updates" might not
> be a
> > good test (wsock2).. I forget if this was win95 or win98....
>
> Ug, you almost gave me a heart attack here... So BOINC_GUI.EXE links against
> WSOCK32.DLL, I think we'll be fine since it'll be up to WSOCK32.DLL which
> winsock provider to talk too.
>
> Although I could be completely wrong here.
>
Just a thought, and I probably should have stated it as so....

The kinder, AZ woody.

Win9x was pretty much 16 bit and did have net problems. Seems some reports of the problem specifically mention "contacting the server" as the root of the problem, and winsock came to mind.. (remember, winsock is only the interface to the driver, and win9x drivers could be long out of date!) (oh heck, what did MS call that where they munged 16bit calls into 32bit call (or vs) in the win9x days?

Maybe some clues here......

> > Isn't the new winsock required for boinc on 98? "no updates" might not
> be a

A list of DLLs, including WSOCK32.DLL, also for win9x can be found:
here

but please fist verify the file size and file date of your WSOCK32.DLL and make a copy of the origial

goodluck to all involved persons

> Win9x was pretty much 16 bit and did have net problems. Seems some reports of
> the problem specifically mention "contacting the server", and winsock came to
> mind.. (remember, winsock is only the interface to the driver, and win9x
> drivers could be long out of date!)
>
> Maybe some clues here......

Yeah, I could see that, except I would have expected the 3.x clients to have run into a problem too.

Stack overflows normally scream out recursive function usage.

I’m hoping Walt can give me a hint as to where in the system this starts, for instance, is in just beginning to construct the scheduler request, sending the request, or parsing the response.

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