PDA

View Full Version : HOWTO: Hoard and DUMP WUs



JPS
01-08-2002, 10:24 PM
-Genome@Home HOWTO #1-
Hoard and Dump WUs using the -nonet flag

It is not uncommon for users to want to run G@H so that they hoard their results for a big ?dump? or submission of results. Likewise, many users run G@H on machines that do not have a direct connection to the net. In both of these cases, it is best to run G@H with the -nonet flag invoked, i.e. C:\Program Files\Genome@home\ghclinet.exe -nonet. When this flag is invoked, G@H simply reprocesses the original WU it has over and over; reseeding itself each time it finishes.

The following HOWTO outlines the simplest and most effective method of hoarding and dumping your results if you use -nonet.

1) Assuming you have installed G@H on your computer(s) and have been crunching for some time using -nonet, the files that are of interest to you now are the .### files [csum.###, output.chi.###, and str.###]. These are the files you want to upload to the Stanford servers to get credit for your work. Depending on how long you have been running ?nonet, it is very possible that you will have hundreds of these in your G@H directory. Note: if you have any .2 files, do not bother with these - they are a cached work unit waiting to be crunched.

2) The simplest way I have found to upload the results is to have a separate, base install of G@H that is used solely to upload work from [you can also use the -upload switch on your current install if you desire - self explanatory]. To get this, I simply do a second install of G@H into a directory separate from the original install, in this case C:\Dump. You can remove the subsequent entries to your start menu if they concern you. You can also copy all of the base files from your original install directory into your C:\Dump directory.

3) Once you have a platform to dump from you need to get all of your hoarded .### files into that directory. Copy all of the stored .### files into the C:\Dump directory but be very careful if you get a warning that the .### extension on the files being pasted in matches the file extension of files already existing in the dump directory - DO NOT PROCEED TO OVERWRITE THE EXISTING FILES WITH THE ONES BEING PASTED INTO THE DUMP DIRECTORY. If you do you will destroy some of your results! What you should do is cancel the paste function and instead, paste them into a separate subdirectory within your dump directory which will be used for files with duplicate .### extensions.

4) You do not need to worry about the id.dat file. If you are missing it, a new one will be generated before you dump. This file simply IDs machines, it does not affect whether or not you get credit for the upload.

5) Make sure your ghclient.cfg file is correct, i.e. it has your user name and our team number: 1881113524. Without this being correct, you will not get credit for your dump.

6) Actually dumping the work is simple: You can run the client executable [ghclient.exe] without any flags by double clicking on the ghclient.exe file, or you can run ghclient.exe via the command line and use the -upload flag. The program will connect to the Stanford servers and upload all of your output files.

7) Now that the main dump is done, you need to upload any .### that had duplicate extensions. Copy those files out of the directory you were storing them in and paste them into your C:\Dump directory. Then proceed as outlined in (6) above.

All done - now don't you feel better???
:D :p :D

FoBoT
01-09-2002, 12:46 PM
were not some of the WU lost in the "great purge" attributed to people hording/dumping ?
isn't it safer to just send as often as possible ?

JPS
01-09-2002, 12:51 PM
Supposedly that is right. The intent of this HOWTO is to outline how to run G@H using -nonet and still get your results posted. This is most often apply to people who have boxes that are connected via dialup or with no connection at all, e.g. I have a T-Bird 1500 sitting in my office at home with non connection what so ever. I load up a WU once a week [taken from a work PC] and let it chomp away...

If you need more info on how to get a machine with no connection to the net up and running G@H - I assumed that had been done in the HOWTO - let me know and I will walk you through it.