Client-core communications error: ERROR 0xc0000005

[Karlsweldt]

Had a bit of a go-round Monday night the 22.
Seems that after submitting a finished WU, a new WU comes down that requires a different core. Core file loaded and the process begun. Then I get the error message that the "F@H process {FahCore_b4.exe} generated errors and will be closed by Windows. An error log is being created". This went on more than 8 times, where I just closed the processes and went to bed. The other process (two running) was OK.
On restarting Tuesday morning, I expected more problems.. but the WU loaded with a new core download, and has had no problems all day!
Curious, though, that Windows did not generate an error listing in the Event Log for the system.. yet the F@H program log had the error listed.
Below is a short summary of the event(s) from the F@H log:



(Machine status)
OS: Microsoft Windows 2000 Service Pack 4
CPU: Intel(R) Pentium(R) 4 CPU 3.20GHz
CPU ID: GenuineIntel Family 15 Model 3 Stepping 4
CPUs: 2 Logical, 1 Physical
Memory: 1023 MB
Threads: Windows
My note: More than 512 MB of free memory available, with two F@H processes running and a Web window open.


Running.. Project: 10005 (Run 1143, Clone 0, Gen 2)

[03:22:30] + Processing work unit
[03:22:30] Core required: FahCore_b4.exe
[03:22:30] Core found.
[03:22:30] Working on Unit 00 [February 23 03:22:30]
[03:22:30] + Working ...
[03:22:34] CoreStatus = C0000005 (-1073741819)
[03:22:34] Client-core communications error: ERROR 0xc0000005
[03:22:34] Deleting current work unit & continuing...
[03:22:38] - Preparing to get new work unit...
[03:22:38] - Presenting message box asking to network.
[03:22:40] + Attempting to get work packet
[03:22:40] - Connecting to assignment server
[03:22:41] - Successful: assigned to (129.74.85.48).
[03:22:41] + News From Folding@Home: Welcome to Folding@Home
[03:22:41] Loaded queue successfully.


Edit: Disabled smiles in this post.

[evasive]

0xc0000005 is a rights issues thingy, windows having a hiccup I think.

[Karlsweldt]

No further problems with that WU process. The other process finished a WU this afternoon, and got a new WU similar to what came down for the first process. There was an error, but all seems to be working OK. Using the same FAH core as the other process.
__________________________________________
Summary: 24/Feb/2010
From running F@H log:

Project: 10007 (Run 1687, Clone 2, Gen 0)
Reading tar file par_all27_prot_lipid.inp
Reading tar file scpismQuartic.inp
Reading tar file ww_exteq_nowater1.pdb
Reading tar file ww_exteq_nowater1.psf
Reading tar file protomol.conf
Reading tar file core.xml
Digital signatures verified
ERROR: Exception in thread 1: @ fah\net\Socket.cpp:139:<unknown> 0: Could not bind socket to 127.0.0.1: No error
________________________________________
Curious, though.. I noted the process was compiled using Win XP. Wonder if the process was formed for a 64-bit OS platform, and is having compatibility problems with the 32-bit platform on my Win 2000 Pro? I am still using the FAT32 format style for compatibility with older programming.

[Pette Broad]

The first error I see a fair bit. It happens when a bad WU is downloaded and because you have not returned anything you get the same WU again (and again). You just have to keep trying and eventually you'll get a different WU.

The second error message "could not bind socket" you only get with Protomol WU's. This happens if you have a dual core or hyperthread and another instance is running. Only one core can bind itself so an error message is generated. It isn't really an error and the WU will process normally.

Pete

[Karlsweldt]

Thanks, Pete..
Have not had any further issues, and have had at least six of this class of WU since. Due in about 4.5 days, I usually finish in 3 days or less. I run about 15~16 hours daily. Machine is shut down for bed time.
I forgot to note in the original post that this class of WU is listed as "ProtoMol".

[Pette Broad]

The Protomol core is still in the experimental stage. Currently, protomol WU's are only being sent to Pentium 4 machines. They, and by that I mean Protomol cores, are capable of using sse3 but there's some sort of bug which causes sse3 capable machines to crash the core to the desktop now and again. So this rules out AMD Athlon 64 and Pentium 5 machines. There's another bug which rules out Socket A machines which don't have sse2 and older Pentium 3 machines have problems with making the deadlines. So you're likely to see quite a few of these until the bugs are ironed out and they can be released to other machines.

Pete

[Karlsweldt]

The only problems with this group of units is as first noted. No problems since. Really have had no problems at all with this system, except long ago when the original memory started acting up. That was PC3200 "value" grade memory! Have installed premium PC3500 memory for peace of mind.
I don't mind being a 'test platform' for new units. No problem meeting deadlines. If I note one with a short submission time, I will just run that and pause the other process until submitted.
Any amount of points added to our group score helps us stay "up there".

[Karlsweldt]

Another instance as first posted. Submitted a regular WU, then when a new one is sent it generates errors.

------------------------------------------------------
Brief:
--------------
+ Processing work unit
Core required: FahCore_b4.exe
Core found.
Working on Unit 09 [March 13 17:12:24]
+ Working ...
CoreStatus = C0000005 (-1073741819)
Client-core communications error: ERROR 0xc0000005
Deleting current work unit & continuing...
- Preparing to get new work unit...
- Presenting message box asking to network.
+ Attempting to get work packet
- Connecting to assignment server
- Successful: assigned to (129.74.85.48).
--------------------------------------------------

This continued for at least 11 instances, where I shut down the other process to see if there was interference. No difference. Shut down this process, ran the other for a few hours.. then tried again. Same result.
Wondering if this server is sending out WUs that aren't intended for my system? Will try again on Mar.14th morning to see if conditions improve to get new WU that is "palatable" to system.

My System Specs:
Chaintech S848P, Socket 478 with Intel 3.20e Prescott/HT. 800 FSB with 1 GB of GeiL CL2-3-3 Ultra-433 (PC3500) memory, 530 watt PSU, nVidia GeForce 6600 AGP 8x 256 Mb video, SB Audigy2 Platinum audio; hard drive is a Seagate 7200 RPM SATA 120 GB. BIOS is dated 02/2004. Running Windows 2000 Pro with SP4. Hard drive has four partitions, all less than 50% full. Task Manager shows more than 400 MB of free memory with two WU processes running. Never goes above 60°C. when working hard.
Has always been stable, very rare with BSOD instances.

[Pette Broad]

Try deleting the b4 core, that'll force a download of the latest core. Could be that your present b4 core is outdated.

Pete

[Karlsweldt]

The F@H process had already taken at least six downloads of the "FahCore_B4" and still would not work. But starting again on Sunday morning, all was normal. Got the 'Project 10005' WU at present. Was never noted as to what WU project was giving the problems.
This is an excerpt from the F@H current log:
-------------------------------------------------
Date: Jan 21 2010
[time mark:] Time: 19:16:23
[time mark:] Revision: 1756
[time mark:] Compiler: Intel(R) C++ MSVC 1500 mode 1110
[time mark:] Options: /TP /nologo /EHsc /wd4297 /wd4103 /wd1786 /arch:IA32 /Ox
[time mark:] /QaxSSE2,SSE3,SSSE3,SSE4.1,SSE4.2 /Qrestrict /MT
[time mark:] Platform: Windows XP
[time mark:] Bits: 32
[time mark:] Mode: Release
---------------------------------------------------

Should really not be that much difference between Windows XP 32-bit and W2K Pro as far as handling sequences.