About Features Downloads Getting Started Documentation Events Support GitHub

Love VuFind®? Consider becoming a financial supporter. Your support helps build a better VuFind®!

Site Tools


Warning: This page has not been updated in over over a year and may be outdated or deprecated.
administration:solr_replication

This is an old revision of the document!


Solr Replication

Solr may be configured to replicate in a master-slave configuration. This can be useful for separating your indexing processes from your running search system, for load balancing, etc. Newer versions of Solr also support a more complex “Solr Cloud” configuration for robust multi-server searching; see VUFIND-834 for some notes.

solrconfig.xml

For basic master-slave Solr replication, the following configuration can be added to the $VUFIND_HOME/solr/$core/conf/solrconfig.xml file (uncomment the appropriate master or slave sections):

  <requestHandler name="/replication" class="solr.ReplicationHandler" >
    <!--
       To enable simple master/slave replication, uncomment one of the
       sections below, depending on whether this solr instance should be
       the "master" or a "slave".  If this instance is a "slave" you will
       also need to fill in the masterUrl to point to a real machine.
    -->
    <!--
       <lst name="master">
         <str name="replicateAfter">commit</str>
         <str name="replicateAfter">startup</str>
         <str name="replicateAfter">optimize</str>
         <str name="confFiles">schema.xml,stopwords.txt</str>
       </lst>
    -->
    <!--
      <lst name="slave">
         <str name="masterUrl">http://master:8080/solr/biblio</str>
         <str name="pollInterval">00:00:60</str>
       </lst>
    -->
  </requestHandler>

pull-browse.sh (AlphaBrowse Synchronization)

A Bash script to pull browse indexes from another host. The idea is that if you are running a Solr slave to keep the indexing load away from searches, you probably want to do the same for the alphabetical browse indexes. But there's no replication for SQLite! Prereqs are curl and passwordless ssh/scp between the hosts for the administrative user.

#!/bin/bash
# Tod Olson <tod@uchicago.edu>
# Much is stolen shamelessly from scripts by Keith Waclena
#
# Copy browse index from a remote host to local host via scp.  Sets up
# the .db-ready file semaphore and calls browse handler to move the
# index into place.
#

ME=`basename $0 .sh`

USAGE="Usage: $ME [-t] [-V remote_home] [user@]host index; -H for help"
HELP="OPTIONS:
 -H: print this help message
 -B: use scp batch and quiet modes
 -V: VUFIND_HOME on remote host (if different from localhost)
 -t: testing only - NYI

ARGUMENTS:
 host:  name of the remote host to copy from, can accept
        SSH-style user@host syntax
 index: name of the index (less the _browse suffix)

ENVIRONMENT
 The following variables must be set:

 VUFIND_HOME"
TAB="$(printf '\t')"

TESTING=false

# prerequisites

prereqs ()
{
   satisfied=
   missing=
   for p in $@
   do
	if type "$p" > /dev/null 2>&1
	then satisfied="$satisfied|  $p${TAB}$(which $p)"
	else
	    missing="$missing  $p"
	fi
   done
   if [ "$missing" != "" ]
   then
       (
	    echo "$ME: missing prerequisites:"
	    echo "$missing"
	    echo -n "$ME: satisfied prerequisites:"
	    echo "$satisfied" | tr \| '\012'
	) >&2
       exit 69
   else : ok
   fi
}

noreq ()
{
   fatal "DRYROT: missing prerequisite!"
}

prereqs scp

# handy functions

usage () {
   echo "$USAGE" 1>&2
   case "$1" in
   '') ;;
   0)  exit 0 ;;
   *)  exit "$1" ;;
   esac
}

nyi () {
   echo "$1: not yet implemented" 1>&2
   exit 2
}

warning () {
   echo "$*" 1>&2
}

fatal () {
   if expr "$1" : '[0-9][0-9]*$' > /dev/null
   then X=$1; shift
   else X=1
   fi
   echo "$*" 1>&2
   exit $X
}

prefix () {
   if $TESTING
   then echo "TEST$1"
   else echo "$1"
   fi
}

#
# Check environment
#
if [ -z "$VUFIND_HOME" ]
then
    fatal "VUFIND_HOME not set"
fi

REMOTE_VUFIND_HOME="$VUFIND_HOME"
SCP_BATCH=""
SSH_QUIET=""

while true
do
   case "$1" in
   -H)     shift; usage; echo "$HELP"; exit 0 ;;
   -t)     shift; TESTING=true ;;
   -B)     shift; SCP_BATCH="-Bq"; SSH_QUIET="-q";;
   -V)     REMOTE_VUFIND_HOME=$2; shift; shift;;
   --)     shift; break ;;
   -)      break ;;
   -*)     usage 1 ;;
   *)      break ;;
   esac
done
[ $# -eq 2 ] || usage 1
REMOTE_HOST=$1
INDEX=$2

#
# Main
#

INDEX_BASE=${INDEX}_browse
REMOTE_BASE=$REMOTE_VUFIND_HOME/solr/alphabetical_browse/${INDEX_BASE}
LOCAL_BASE=$VUFIND_HOME/solr/alphabetical_browse/${INDEX_BASE}

# .db-ready is the signal that .db-updated is complete, but
# .db-updated could move into place and .db-ready could be deleted at
# any time
if ssh $SSH_QUIET $REMOTE_HOST test -f ${REMOTE_BASE}.db-ready
then
    scp ${SCP_BATCH} -p ${REMOTE_HOST}:${REMOTE_BASE}.db-updated ${LOCAL_BASE}.db-updated ||
    scp ${SCP_BATCH} -p ${REMOTE_HOST}:${REMOTE_BASE}.db ${LOCAL_BASE}.db-updated
else
    scp ${SCP_BATCH} -p ${REMOTE_HOST}:${REMOTE_BASE}.db ${LOCAL_BASE}.db-updated
fi

if [ $? -ne 0 ]
then
    fatal "failed to copy $INDEX_BASE from $REMOTE_HOST"
fi

touch ${LOCAL_BASE}.db-ready &&
curl "http://localhost:8080/solr/biblio/browse?source=${INDEX}" > /dev/null

# Local Variables:
# End:
administration/solr_replication.1450120060.txt.gz · Last modified: 2015/12/14 19:07 by demiankatz