Corrupting WU?

[Leoslocks]

Project: 2665 (Run 2, Clone 769, Gen 10)

Project: 2653 (Run 29, Clone 28, Gen 71)

I am having some network issues with one SMP client.
Q9450 @ 3.2GHz
Windows Vista Ultimate 64
2X 2Gb ram

The two WUs were stuck in que and would not send. I ran qfix and the units were sent.

Code: Select all

# SMP Client ##################################################################
###############################################################################

                       Folding@Home Client Version 5.91beta6

                          http://folding.stanford.edu

###############################################################################
###############################################################################

Launch directory: C:\SMP6
Executable: C:\SMP6\fah.exe


[03:46:21] - Ask before connecting: No
[03:46:21] - User name: Leoslocks (Team 34878)
[03:46:21] - User ID: 1BD42BDD2D56C959
[03:46:21] - Machine ID: 6
[03:46:21] 
[03:46:21] Loaded queue successfully.


[03:46:21] 
[03:46:21] + Attempting to send results
[03:46:21] + Processing work unit
[03:46:21] Core required: FahCore_a1.exe
[03:46:21] Core found.
[03:46:21] Working on Unit 08 [June 17 03:46:21]
[03:46:21] + Working ...
[03:46:21] 
[03:46:21] *------------------------------*
[03:46:21] Folding@Home Gromacs SMP Core
[03:46:21] Version 1.74 (March 10, 2007)
[03:46:21] 
[03:46:21] Preparing to commence simulation
[03:46:21] - Ensuring status. Please wait.
[03:46:38] - Looking at optimizations...
[03:46:38] - Working with standard loops on this execution.
[03:46:38] Examination of work files indicates 8 consecutive improper terminations of core.
[03:46:49] - Expanded 4710046 -> 24426905 (decompressed 518.6 percent)
[03:46:51] 
[03:46:51] Project: 2665 (Run 0, Clone 889, Gen 12)
[03:46:51] 
[03:46:53] Entering M.D.
[03:46:59] Calling FAH init
[03:47:00] Read topology
[03:47:00] (Starting from checkpoint)
[03:47:01] ut of 250000 steps  (3 percent)
[03:47:01] water
[03:47:01] Writing local files
[03:47:01] Completed 7500 out of 250000 steps  (3 percent)
[03:47:06] Extra SSE boost OK.
[03:53:05] + Results successfully sent
[03:53:05] Thank you for your contribution to Folding@Home.
[03:53:05] + Number of Units Completed: 6



[03:53:05] + Attempting to send results
[03:55:05] + Results successfully sent
[03:55:05] Thank you for your contribution to Folding@Home.
[03:55:05] + Number of Units Completed: 7

Just wondering if this setup is corrupting the WUs.

[7im]

Vista is better about not instantely erroring out when there is a network change/drop (no 0x7b errors), but it can still cause the WU to hang at 100% and not upload. I have seen several reports about that.

How are you connected to the internet? DSL, Cable, etc? How are you connected to your internet hardware? Cat5, USB, WIFI, WIMAX, etc?

[bruce]

Do you have BigWU enabled? During the early parts of the SMP beta, there was a recommendation that all SMP WUs were big so you needed to configure that setting to get work. I suspect that since almost everybody has that setting enabled, reports of such problems have not reached the project owners and they may not have configured all the server exclusions to avoid such problems.

The log you posted doesn't help us debug the problem because it doesn't show a problem. I do know that if you are assigned a WU that creates a results file that is bigger than the setting you have configured, you'll have trouble uploading but qfix will correct this inconsistency. Those symptoms fit your problem description, but that doesn't prove that I've guessed the exact cause of your problem.

Can you give us the PRCG numbers for those WUs? {From qd or from FahLog(-Prev).txt}

[Leoslocks]

Project: 2665 (Run 2, Clone 769, Gen 10)

Project: 2653 (Run 29, Clone 28, Gen 71)

I have completed quite a few WUs on this machine but it is the newest crop in my garden. I am connected Via Comcast Cable > Linksys Router > Cat5e cable to a Switch (31 meters)> 6 machines on the switch Cat5 hardwired with no cable longer than 3 meters.

I posted in this forum to determine if the WU's may have gotten corrupted. I know I need to address the networking on this particular machine. The onboard nic shows one yellow light steady. I have internet access and have downloaded files to this machine. Windows updates have been applied to it also(170 meg initial update).

I expect the client to hang if I adjust the network hardware/software. Ctrl + C and restarting the client usually sends the results and continues folding. These units would not send but the client kept folding.

[bruce]

The one-light / two-light generally depends on whether you're running at 10MB or 100MB, but it may depend on who makes the NIC.

Win-SMP 5.91 and 5.92 respond quite differently to network changes: V5.92 resumes work after an error; v5.91 deletes the work.

As a general rule, changing the network configuration from DHCP/Automatic-IP to preassigned IP addresses avoids 99% of the network changes.

[Leoslocks]

I just tried a different network cable. Two green lights on that port for the switch. One Orange steady light on the client nic with the second blinking yellow so I assume it is working correctly. I adjusted some setting for the nic. Something about Jumbo frames....
Static IP is and has been assigned to the nic.