Creating New Uploaded Corpora Users

TODO: this is obsolete and should be ignored! We're no longer using client/server mode for uploaded-corpora mass-checks....

These are instructions for PMC members who wish to set up a new account for users to upload their HandClassifiedCorpora for us to use in the NightlyMassCheck runs.

This method uses mass-check's client-server mode to distribute the load. However this seems to be occasionally buggy. If the resulting mass-checks run into problems, alternatively see UploadedCorporaIndependentMassCheck to set up a mass-check uid in traditional, non-client-server mode.

Log in to the zone and run:

MCUSER=[username]
MCPWD=[random password]

sudo mkdir /export/home/nitemc/$MCUSER
sudo chmod 1777 /export/home/nitemc/$MCUSER
cd /export/home/nitemc/$MCUSER
echo "$MCPWD" > rsync_password
chmod 600 rsync_password

sed -e "s/MCUSER/$MCUSER/" -e "s/MCPWD/$MCPWD/" > .corpus

And paste in these lines:

opts_weekly="--net -j 8 --reuse --cache --cachedir=/tmpfs/aicache_nightly --cs_schedule_cache --cs_cachedir=/export/home/nitemc/cache --restart=500 ham:detect:/export/home/bbmass/uploadedcorpora/MCUSER/ham/* --after="-15552000" --tail=25000 spam:detect:/export/home/bbmass/uploadedcorpora/MCUSER/spam/*"
opts_nightly=" --reuse --cache --cachedir=/tmpfs/aicache_nightly --cs_schedule_cache --cs_cachedir=/export/home/nitemc/cache --restart=500 ham:detect:/export/home/bbmass/uploadedcorpora/MCUSER/ham/* --after="-15552000" --tail=25000 spam:detect:/export/home/bbmass/uploadedcorpora/MCUSER/spam/*"
tmp=$HOME/tmp
tree=$HOME/svn
prefs_weekly=$HOME/user_prefs.weekly
prefs_nightly=$HOME/user_prefs.nightly
username=bb-MCUSER
password=__RSYNC_PASSWORD__
serverhost=spamassassin.zones.apache.org.:38899
clienthosts=__CLIENTHOSTS__
clienttree=nightlymc_MCUSER

Then CTRL-D to end cat.

mkdir tmp
svn co http://svn.apache.org/repos/asf/spamassassin/trunk svn
[accept certificate 'p'ermanently]

sudo chown -R nitemc .

In SVN trunk, edit build/nightlymc/run_nitemc, add their username to the list, check that file in.

Then in the zone, as the uid "automc", do this:

  cd /home/automc/svn/spamassassin
  svn up

so that that latest script is updated for when cron runs.

Finally, edit /home/corpus-rsync/secrets and add a line to the end, like so:

bb-$MCUSER:$MCPWD

e.g. if MCUSER was "jm" and the generated MCPWD was "Wi0FdPWg":

bb-jm:Wi0FdPWg

That's it. Now tell the user their username (with the "bb-" prefix) and password.

NewUploadedCorporaUser (last edited 2009-09-20 23:17:21 by localhost)