Stack overflow error when running FaD

[Fardringle]

I've been working on getting my remote 'herd' up and running again today. Most of them went flawlessly, but a few of them got an odd Stack Overflow error when starting up the client. I managed to get all but one of them fixed by running FADSETUP.EXE again on the problem clients. However, one of them keeps getting the same error regardless of what I do when I have the client start when Windows starts. If I manually browse to the folder on the server where the FaD files are located, I can start the client normally from there, but it gets the following error if started from the registry, from the Start Menu, or by a scheduled task:

Salford run-time library.
Stack Overflow: Re-link program with bigger stack value.
(stack:reserve,commit)
. Will attempt to trace back.

I need to leave the office and head back home very shortly so I'd highly appreciate any ideas you can give me before I go... 🙂

If I don't get an answer shortly, I'll just have to disable the client on this PC for a while, but I'd rather not have to do that. 😉



EDIT: Well, all of the clients were running nicely until they started returning results about 2 hours ago. Since that time, every result returned is showing as "Rejected" in the server's queue window, and all of the clients that have been rejected now refuse to run THINK.EXE normally. There are plenty of new jobs available in the server's queue and I even manually downloaded a dozen more just to be sure (there aren't any old/expired jobs in the queue).

Any time I try to stop and restart the program on the 'rejected' computers I get an entry like this in the log:

04-SEP-05 06:13 Loader started in F:\Program Files\Find-a-Drug\ on KAREN
04-SEP-05 06:13 Local copies in C:\DOCUME~1\karen.TAM\LOCALS~1\Temp
04-SEP-05 06:13 Skipped running : THINK
04-SEP-05 06:13 Startup : fadsetup
04-SEP-05 06:13 Loader finished on KAREN

I checked and there are copies of THINK.EXE and FADSETUP.EXE in the Temp folder (which has never happened before that I am aware of), but even if I delete those files and try to launch the client again the local copies show up again and I get another of the same entry in the log.

I have confirmed that everyone on the domain has Full Control access to the Find-a-Drug folder and all of its contents and subfolders on the server so I don't know what else could be causing this.

Any ideas what the problem might be?

Thanks,
Fardringle


additional update: To make things even stranger, the two AMD machines at the site and the four processes running on the dual P4 Xeons on the server are all still crunching and happily returning results with no problems at all. But ALL of the P4 (and Celeron) workstations in the office had their results rejected and will not restart the THINK client...


Yet another (and hopefully final) update I replaced a corrupt .DLL file and corrected some missing permissions on the FaD share (thanks to the very wise advise of Mondobyte) and it looks like all of my machines are back up and operational again. Yay! 😀

 

[mondobyte]

Send me a PM ... I can help ... "unofficially, of course"

Thanks ... mondo

 

[Fardringle]

There may still be a software problem, but I managed to find a 'fix' at least temporarily.

Rather than having the tasks run the project from the mapped drive like it does on all of the other machines, I set it to run from the UNC name (full network path) where the files are located and the problem went away. I suspect it's a problem with the way this one station is connecting to the new (really neat) server I installed yesterday but I ran out of time to troubleshoot it for now.

If you do know of a permanent fix, I'd like to have it. Or I might just set all of the workstations to run the service from the UNC name instead of the share since that will let them run before login (before the drive gets mapped in the login script). 🙂


PMs responded to, by the way.. 😉

 

[Fardringle]

Well, FaDSpy says that everything is running nicely so far at the office. Most of the machines there will only be on part time, and the only Intel machine running Hyperthreading is the INFINITY (server) machine - and I might have to knock it down to one process per CPU instead of two (total of 4) if it has any impact on the server operation. Add these on top of my two home machines (XP 2000 running at 1730Mhz, and Barton 2600 running at 2Ghz) as well as my office machine (P4 3.2Ghz on most of the time) and two remote machines that run about 2 hours per day and I should be back up to at least 40-45K points per day again. 😀


p.s. This is a snapshot of the systems I brought back online today after a little more than a month of down-time due to contractor stupidity:

System Name.....System.....CPU.............CPU Speed.......OS
Darrin................Intel.........1700 Mhz.....(1x1700Mhz).....Windows NT/2k/XP
Chris.................Intel.........2000 Mhz.....(1x2000Mhz).....Windows NT/2k/XP
Ward.................AMD.........1463 Mhz.....(1x1463Mhz).....Windows NT/2k/XP
Kari...................Intel.........2800 Mhz.....(1x2800Mhz).....Windows NT/2k/XP
GinaMarie........AMD.........2015 Mhz.....(1x2015Mhz).....Windows NT/2k/XP
John..................Intel.........2000 Mhz.....(1x2000Mhz).....Windows NT/2k/XP
Karen................Intel.........2660 Mhz.....(1x2660Mhz).....Windows NT/2k/XP
Donna...............Intel.........2660 Mhz.....(1x2660Mhz).....Windows NT/2k/XP
Barb..................Intel.........3000 Mhz.....(1x3000Mhz).....Windows NT/2k/XP
Steve.................Intel.........3000 Mhz.....(1x3000Mhz).....Windows NT/2k/XP
Infinity................Intel.........6400 Mhz.....(2x3200Mhz).....Windows NT/2k/XP
Jason................Intel.........2000 Mhz.....(1x2000Mhz).....Windows NT/2k/XP
Lisa...................Intel.........2800 Mhz.....(1x2800Mhz).....Windows NT/2k/XP
Paula................Intel.........1700 Mhz.....(1x1700Mhz).....Windows NT/2k/XP
Ryan.................Intel.........2000 Mhz.....(1x2000Mhz).....Windows NT/2k/XP
Dave.................Intel.........1700 Mhz.....(1x1700Mhz).....Windows NT/2k/XP
Starr..................Intel.........3000 Mhz.....(1x3000Mhz).....Windows NT/2k/XP
Rae...................Intel.........3000 Mhz.....(1x3000Mhz).....Windows NT/2k/XP

 

[Fardringle]

Bump for a new problem even more annoying than the original (details in the top post).

 

[mondobyte]

I like your sig Fardringle 😉

 

[Fardringle]

Hehe, thanks. 🙂

It's been a "work in progress" since I first started running SETI in October of 2000.

I haven't felt very creative lately, and I've been trying to come up with a better one for FaD since I don't really like the Elvis "quote", but I've made sure to add a line to it for each project that I ran seriously (i.e. for more than a week). The Al Gore "quote" isn't for a specific project, but I added that one when the media was rabidly mis-quoting Gore by taking his statement that he was part of the project that set up the original University/Government "internet" (which is true, by the way - he was part of that project) and making it look like he was claiming that he invented the Internet all by himself. Even though it's not for a specific project, I liked that one far too much to ever consider taking it out. 😀

I have a feeling the bottom line may be the last new one for a while... 😉


p.s. I really appreciate your PM'd suggestions for fixing the problems at the office. I'll let you know after I have a chance to try them out in the morning.

 

[Fardringle]

OK, you've piqued my creativity.

"Borrowing" heavily from the last paragraph of this article that CBS wrote about religion in the U.S.:

Critics call it a FaD. Believers call it crunching to Find-a-Drug. For Fardringle and his friends, it's about finding a cure. - CBS News

It's still not "perfect" but I think it's a whole lot better than the lame Elvis line. 😛

edit: I like the line, but it's probably too obscure of a quote to use without linking to the original. I'm definitely open to suggestions!