PDA

View Full Version : Any way to force 'send' ?


BDR
22 Oct 2003, 12:35pm
I've had enough of trying to fold on the Celi. Is there any way to send in what it has done up to now?
-send x doesn't seem to be working.

Celi gets operated on today... Red Hat gets installed..

Enverex
22 Oct 2003, 12:46pm
Do you mean send in an incomplete WU? If you do then no, because no-body cares about incomplete work units.

If you mean saved WUs then just closing and opening the program should send the WU.

NS

BDR
22 Oct 2003, 12:59pm
Yep, that's what I meant.

I know that sometimes partialy fininshed wu's are sent, but I guess that's for "early end" only, and no way to force a wu to end either?

mmonnin
22 Oct 2003, 02:08pm
Nope, you cant do that.

-send all will work for completed WUs.

BDR
22 Oct 2003, 02:11pm
Ok, thanks.

I hated to lose the wu, but too late now.
Drive has been wiped and RH is being installed.

muddocktor
22 Oct 2003, 04:13pm
OH well, too late for this one but you can always copy a WU over to another rig and finish it there. You copy the work folder(obviously) and also the queue.dat file and fahlog file and you can finish it on the other machine or even the same machine after a format/reinstall of the same base OS, ie Windows(all flavors), Linux, Mac.

BDR
22 Oct 2003, 04:17pm
Shoot.. I didn't think of that. It would have been easy too, since the pc's were all networked. :bawling:

muddocktor
22 Oct 2003, 04:21pm
Yeah, you could have pulled it right over the lan. I've learned a bit about copying work from rig to rig while working offshore. I was on a job earlier this year where I couldn't get my lappy on the net at work but I had net access via a machine on that rig. I sneakernetted about 250-300 points worth of WU's that hitch at work.:cool:

edcentric
22 Oct 2003, 05:22pm
and mudd can be so sneakey.
Maybe you should write up a little step-by-step so that people can refer to it when the need to salvage a wu.

BDR
22 Oct 2003, 05:36pm
edcentric had this to say

Maybe you should write up a little step-by-step so that people can refer to it when the need to salvage a wu.

yep, it would have saved me 32 points.

muddocktor
23 Oct 2003, 01:58am
Yeah, I guess I could write up a little "how to" on it but it is really very easy. The situation that BDR was in is the reason why Stanford included the ability for the client to be able to finish work started on another machine.:cool:

To finish a WU off on a different machine than the one it started from, you will have to copy the work folder and the queue.dat file and fahlog.txt file into the recipient machine's folder where you are running F@H. Be sure that this machine has finished it's work before replacing it's work folder and queue.dat file before transferring it in. Then, you should be able to start the client and it will finish and send the WU. Be aware that the work folder alone can be well over the size that a floppy disk can hold, so if you are using floppies, you will need a couple of them. A pen drive would be the ideal media to transfer the WU.

If you want to sneakernet some work, you can use the same basic system to transfer the work from the net-enabled rig to the no-net rig and back again. When the no-net rig has finished the WU, shut the client down and transfer the entire work folder and queue.dat file and I always transfer the fahlog.txt file too. The WU's result's will be between 1 to over 2 MB, so it makes it harder to transfer the work by floppy, but not impossible. I've taken the work folder and zipped it to floppy, which will span it across 2 floppies if necessary and then reinflated it on the machine with net access with no problems.

Another trick you can do if you are going somewhere with a laptop and have no net access but still want to fold with it is to set up 4 instances of the client and give them machine ID's from 1 to 4. Then, right before you leave, go download some work into all 4 folders. You can either leave all 4 instances running at the same time or just run one at a time and shut down the client when it finishes the WU. That is the way I prefer to do this as you have less chance of something happening and borking up all 4 WUs.

One other thing is that Stanford only gives the credit for the transferred WU to the first machine to turn it in, so that nobody can cheat.;)