SolrCloud backup/restore without shared disk

solrcloud-backup-and-restore.sh

#!/usr/bin/env bash

# https://lucene.apache.org/solr/guide/7_5/making-and-restoring-backups.html explains that 
# SolrCloud Backup/Restore requires a shared file system mounted at the same path on all nodes.
#
# This script uses the old Solr Backup/Restore procedure, which is useful if you don't have
# access to a shared disk.
#
# Requires that the target collection is already set up, with identical number of shards/replicas.

set -o errexit -o pipefail -o nounset

COLLECTION=my_collection_name
FROMHOST=old.solr.host:12655
TOHOST=new.solr.host:12599
MYUSER=myuser

SSHOPTS="-o StrictHostKeyChecking=no"
TMPDIR=/tmp/backup_$COLLECTION

DRY=dry
#DRY=

# Should produce a list of "replica host port"
function getstat() {
  curl -s "http://$1/solr/admin/collections?action=CLUSTERSTATUS&wt=json" \
    | grep -e $COLLECTION -e \"base_url\" \
    | grep -A1 ${COLLECTION}_shard \
    | cut -d\" -f4 \
    | perl -0777 -pe 's|(\d)\n|\1 |g; s|http://||g; s|:(\d+)/solr| \1|g' \
    | grep $COLLECTION > $2
}

function dry() {
  echo WOULD HAVE RUN: $@
}

# oldreplica oldhost oldport newreplica newhost newport
function backup() {
  $DRY ssh -n $SSHOPTS $MYUSER@$2 mkdir -p $TMPDIR/$1
  $DRY curl -s "http://$2:$3/solr/$1/replication?command=backup&numberToKeep=1&location=$TMPDIR/$1"
}

# oldreplica oldhost oldport newreplica newhost newport
function restore() {
  $DRY ssh -n $SSHOPTS $MYUSER@$5 mkdir -p $TMPDIR
  $DRY ssh -n $SSHOPTS $MYUSER@$2 "cd $TMPDIR && rsync -e 'ssh $SSHOPTS' -av $1 $MYUSER@$5:$TMPDIR/"
  $DRY curl -s "http://$5:$6/solr/$4/replication?command=restore&location=$TMPDIR/$1"
}

getstat $FROMHOST /tmp/old
getstat $TOHOST /tmp/new

paste -d" " /tmp/old /tmp/new > /tmp/pasted

# Format of /tmp/pasted:
# oldreplica oldhost oldport newreplica newhost newport

printf '\nSTATUS:\n'

cat /tmp/pasted

printf '\nPress any key to backup\n'
read foo

printf '\nBACKUP\n'

cat /tmp/pasted | while read i; do
  backup $i
done

printf "\nDone. You should really run http://${FROMHOST}/solr/INSERT_REPLICA_NAME/replication?command=details on all backup hosts to check backup states before restoring\n"
printf '\nPress any key to restore\n'
read foo
printf '\nRESTORING\n'

cat /tmp/pasted | while read i; do
  restore $i
done