Hello Janos,
just to let you know: after rebuilding the sphinx index, the problem was
solved.
If someone is interested, I made this crude little script to do the
reindexing in batches (saved as /usr/local/bin/piler_reindexer.sh):
#!/bin/bash
# before reindexing:
# disable cron jobs for piler
# stop searchd: /etc/init.d/rc.searchd stop
# delete sphinx index: rm -rf /var/piler/sphinx/*
# rebuild index: reindex --all
# start searchd: /etc/init.d/rc.searchd start
# run this script: /usr/local/bin/piler_reindexer.sh 2000
# reenable cron jobs
from=1
step=$1
from=`mysql <pilerDB> -u <pilerUser> -p<password> \
-e "select count(*) from metadata" | grep "[0-9]"`
echo "run from $from to $max step $step"
while [ $from -gt 1 ]
do
to=$(($from - $step + 1))
if [ $to -lt 1 ]
then
to=1
fi
echo "import from $from to $to"
reindex -f $to -t $from
/usr/local/libexec/piler/indexer.delta.sh
/usr/local/libexec/piler/indexer.main.sh
from=$(($to - 1))
done
(worked fine for me, but only use at your own risk)
Best regards,
Joern
