El Martes 16 Jun 2009 11:28:07 Milo Thurston escribió: > > and the scheduler, validator and assimilator can see files uploaded > > via the other server, > > Do you mind if I ask for clarification of what you mean in this case?
The "normal" way things work is: When a client reports a task, the scheduler needs to confirm that it already uploaded the output files. Then the validator needs to check if the uploaded files are valid, and, if there is replication, compare them to other results from the same workunit. Finally the assimilator has to read the files to make them do something useful for the project, like putting the results into a database. A lot of this may be different for CPDN, because of trickles. I obviously have no idea how *your* backend stuff works ;) No matter what server handles the file upload, the other daemons need access to the uploaded files. I don't really know how things have to be setup if you have more than one upload server. I guess it would easily work if all write to the same filesystem (which of course may be in a huge RAID of many disks). _______________________________________________ boinc_dev mailing list boinc_dev@ssl.berkeley.edu http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
