New SMP woes

[waltherone]

Finally put linux on my C2D e4300 box and it completed it's first WU, but then after that all hell broke loose. Now it won't do anything, it just keeps giving me variations of this in the log:

[16:53:54] - Starting from initial Work Packet
[16:53:54]
[16:53:54] Project: 3027 (Run 8, Clone 95, Gen 11)
[16:53:54]
[16:53:54]Entering M.D.
NNODES=4, MYRANK=3, HOSTNAME=waltherone-desktop
NNODES=4, MYRANK=0, HOSTNAME=waltherone-desktop
NNODES=4, MYRANK=1, HOSTNAME=waltherone-desktop
NNODES=4, MYRANK=2, HOSTNAME=waltherone-desktop
[cli_3]: aborting job:
Fatal error in MPI_Wait: Error message texts are not available
[16:54:01] Finalizing output
[cli_3]: aborting job:
Fatal error in MPI_Wait: Error message texts are not available
[cli_3]: aborting job:
Fatal error in MPI_Wait: Error message texts are not available
[16:54:05]
[CoreStatus = 66(102)
[16:54:05] + Shutdown requestd by user. Exiting.

I didn't request to shutdown the program btw. Sometimes in place of that it will say deleting current work unit and continuing.. then DL another WU, and get the same error message. Sometimes it'll say Entering M.D. as above, and then the very next line it'll reprint all of the WU info (project/run/clone/gen) and Entering M.D. again. Like it's losing it's place or something

What gives?

Already tried redownloading and installing. Manually deleted everything as well, no dice.

 

[GameALot]

What version of Linux are you using? I have Ubuntu v6.1 on my Asus P5B-E with an E-6300 and have had few problems.

The only thing I can recommend is lowering your OC (blasphemy I know ), and trying again.

 

[waltherone]

It's ubuntu 6.10 with the 32-bit libs added that SMP needs to run. Ran fine for 12hrs during it's first WU, just after that it refuses. I guess I'll set the bitch back to stock and try that... blasphemy indeed

EDIT: No luck

 

[GameALot]

Have you checked the system log for errors that happened at the time of the crash?

Is your loop-back device set to 127.0.0.1? Both the localhost and hostname on my system are assigned to 127.0.0.1.

I think your system is having trouble syncing processes. None of my logs have the part about NNODES in them.

 

[Tigerbiten]

My clients show the 4 lines of NNODES but they point to "localhost.localdomain" not "waltherone-desktop".
I second it could be a network-hosting problem.
Try pinging "waltherone-desktop" and see what you get.
Ideas came from .... http://forum.folding-community.org/ftopic17985.html

Luck ............

 

[waltherone]

Tiger, I think you were on to something. Last night AFTER I started the SMP client running I started messing with settings (mind you last night was the first time I ever even saw a linux interface, ever) to see if I could find a way to share the folding folder with my windows network so I could let EMIII monitor it (is that even possible with SMP?). Maybe something I changed along the way messed it up, but I never was able to figure out what that was. In the end, I did what most linux guru's shudder at the thought of, and just reinstalled it. It's not like I had it "customized" as I installed it last night anyway, so I just reinstalled, and it's working fine now on a big project (2605). Once it finishes (tomorrow afternoon I think) I guess we'll see if it is able to continue this time

All I can say is, this e4300 @ 3ghz running SMP has the potential for some nice PPD increases, especialy after being crippled by all those damn super_villians and p2124/2125's

 

[Tigerbiten]

Re-install is my nornal answer after I've broken my Linux install as well.
My next trick is to get samba running so I can monitor them aswell.
But they are running and that will do so far ........

Luck ...........

 

[zackboll]

make sure to remember to run the command smbpasswd after you get your samba server running