VMware (linux) Client Help

ChrissyT88

ChrissyT88

ChrissyT88

Associate
Joined
9 Mar 2008
Posts
1,039
Hi all - i set up the linux client (on Ubuntu) in VMware in the last few days and it has been crunching away happily, and tonight finished its first WU. However, i have had a few problems... The WU finished, and the client shutdown, but the terminal window showed a message along the lines of 'no access to mpich' - i think (unforutnately i did not keep the screen open to write it down, and FAHmon and the log files do not seem to have recorded the error). The log file reads as follows:

Code: 
[23:20:16] Completed 247500 out of 250000 steps  (99%)
[23:30:51] Completed 250000 out of 250000 steps  (100%)
[23:31:52] 
[23:31:52] Finished Work Unit:
[23:31:52] - Reading up to 21141072 from "work/wudata_01.trr": Read 21141072
[23:31:54] trr file hash check passed.
[23:31:54] - Reading up to 27623324 from "work/wudata_01.xtc": Read 27623324
[23:31:55] xtc file hash check passed.
[23:31:55] edr file hash check passed.
[23:31:55] logfile size: 181538
[23:31:55] Leaving Run
[23:31:59] - Writing 49176798 bytes of core data to disk...
[23:31:59]   ... Done.
[23:32:09] - Shutting down core
[23:32:09] 
[23:32:09] Folding@home Core Shutdown: FINISHED_UNIT
[23:34:12] ***** Got an Activate signal (2)
[23:34:12] Killing all core threads

Folding@Home Client Shutdown.


--- Opening Log file [October 31 23:34:20] 


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

                       Folding@Home Client Version 6.02

                          http://folding.stanford.edu

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

Launch directory: /home/chris/folding
Executable: ./fah6
Arguments: -smp -verbosity 9 -oneunit 

[23:34:20] - Ask before connecting: No
[23:34:20] - User name: ChrissyT88 (Team 10)
[23:34:20] - User ID: 6863B6A61C0C450E
[23:34:20] - Machine ID: 1
[23:34:20] 
[23:34:20] Loaded queue successfully.
[23:34:20] - Autosending finished units...
[23:34:20] Trying to send all finished work units
[23:34:20] + No unsent completed units remaining.
[23:34:20] - Autosend completed
[23:34:20] 
[23:34:20] + Processing work unit
[23:34:20] Core required: FahCore_a2.exe
[23:34:20] Core found.
[23:34:20] Working on Unit 01 [October 31 23:34:20]
[23:34:20] + Working ...
[23:34:20] - Calling './mpiexec -np 4 -host 127.0.0.1 ./FahCore_a2.exe -dir work/ -suffix 01 -checkpoint 15 -verbose -lifeline 6449 -version 602'

[23:34:20] 
[23:34:20] *------------------------------*
[23:34:20] Folding@Home Gromacs SMP Core
[23:34:20] Version 2.01 (Wed Aug 13 13:11:25 PDT 2008)
[23:34:20] 
[23:34:20] Preparing to commence simulation
[23:34:20] - Ensuring status. Please wait.
[23:34:20] Files status OK
[23:34:20] 
[23:34:20] Project: 0 (Run 0, Clone 0, Gen 0)
[23:34:20] 
[23:34:20] Error: Could not write local file.  Exiting.
[23:34:20] - Shutting down core
[23:34:30] one 0, Gen 0)
[23:34:30] 
[23:34:30] Error: Could not write local file.  Exiting.
[23:34:30] - Shutting down core

The client now does not seem to be able to do anything (keeps repeating the bottom section of the code), which is a pain. I am loathed to delete the work folder etc since i have no idea if it sent the finished unit, or how i can solve this problem if it happens again. Any help would be greatly appreciated - im pretty new to this VMware thing!
 
Last edited:
Thanks senture - ran qfix and everything seems fine (well, at leats the old unit was sent). Thanks for the shutdown script advice - i will look into it today.

Thanks again
 
Well, i have been trying to run the winSMP client at the weekends (really, its the only time its on lots), but unfortunately it has given me nothing but problems, so i thought i would try the linux smp client, which seems altogether a lot better! Not really sure i could afford another gpu - i bought a GX2 (partially for gaming, mostly for folding!), so power sonsumption is pretty high, and im not sure its (morally... although...) a good idea to burn more electricity, and eat through me and my flatmates money. If i didnt have to sleep (or pay the leccy bill), i could (and would) be churning out 12.5K from this machine. But for now, I think your safe where you are...
 
Good idea! I wish i had been into folding a few years ago when i was in halls - nice free electricity! To be honest, im sure it would only be a few pounds a week, even if it were on 24/7, but seeing the figures for a years usage etc is quite scary!
 
Nothing to do with linux, but saves a new thread. I have been manually setting the affinities of all the fah_core11s and the VMware executable in order to try and even out cpu core temperatures and to localise the vmware client onto two cores that share the same L2 cache. The only problem is that fah_core11 closes after every gpu WU has been completed, and starts again with a different affinity, and sits there doing nothing until i manually change the affinity (over the last two days, i reckon i have lost ~3hrs of gpu time). I have tried some affinity changing programs (set afinity II, prifinitty - both seem pretty good, but not quite what i am after), so have resorted to some batch files. They seem to work, but i've only ever written about 2 batch files ever, so wanted to check to make sure nothing was going to go wrong. Of course, if anyone wants to use the same scripts, feel free.

Code: 
@echo off
echo Starting FAH GPU1
echo -------------------------------
echo.
set NV_FAH_CPU_AFFINITY=2
"C:\Folding@Home\GPU1\[email protected]"

I turned the 'affinity lock' variable off in the client config (and added parameters -gpu x -verbosity 9), and as i say, it seems to run. Just want to check to make sure it wont do any damage...

EDIT: I should add, to quit, im just using ctrl+c, and telling the batch script to end...
 
I just downloaded the VMware workstation trial to see if it was much different, and im amazed at how much better it is than the free server version! Its so much smoother/generally quicker, and doesnt seem to be using any additional cpu resources. It almost makes me want to shell out and pay for it... if it wasnt $189! :eek:
 
I agree - i have tried them both, and virtual box is much better than the free version of vmware, and feels pretty smooth like the paid for version of VMware. The reason i have not been using virtualbox is that it lacks multi 'virtual' processor support, which i thought was required for the SMP client. Can you run an SMP client on a virtualbox virtual machine? If so, i'll be ditching VMware when this free trial runs out!
 
You must log in or register to reply here.
Back
Top Bottom