Create (if necessary) and replicate all databases from a couchdb server to another one

couchclone.sh

#!/bin/sh
#
# Janos Kasza (@janoskk)
#
# Creates (if necessary) and replicates all databases from a couchdb server to another one
#

if [ -z "$2" ]; then 
    cat <<EOF 
Usage: $0 <sourceUrl> <targetUrl>
Creates (if necessary) and replicates all databases from a couchdb server to another one. 
Note that the urls must not contain the trailing '/' and the targetUrl may need to contain 
authentication for admin.

Example: $0 http://example.com:5984 http://admin:admin@localhost:5984
EOF
    exit 1
fi 

DBNAMES=`curl -s -X GET $1/_all_dbs | sed 's/\[//;s/\]//;s/"_[a-z]*"//g;s/[,\"]/ /g'`

for i in $DBNAMES; do 
    curl -s -X PUT $2/$i
    curl -s -X POST $2/_replicate  -d '{"source":"'$1/$i'", "target":"'$2/$i'"}' -H "Content-Type: application/json"
done

I just wanted to say that this script is awesome, but it filters any databases that begin with an underscore. If you want to replicate system databases like _users, replace line 20 with:

DBNAMES=`curl -s -X GET $1/_all_dbs | sed 's/\[//;s/\]//;s/[,\"]/ /g'`

Note that I don't have any database like _replicator to worry about replicating -- I just want _users (I use CouchDB 2.0). Make sure you want to replicate all of these if you have them before you make this change...

Nice little script. If you want replication to be continuous and restartable (i.e. continue if you restart the couchdb sever) then change line 24 to:
curl -s -X POST $2/_replicator -d '{"_id" : "'$i'", "source":"'$1/$i'", "target":"'$2/$i'", "continuous": true}' -H "Content-Type: application/json"
The _id of the record in the _replicator table will be the table name it is replicating. Tested with CouchDB 1.6.1

This is a great script and the way that I used to replicate a cluster of databases. I needed something a little more robust that supports concurrency and is fault-tolerant, so I created replicate-couchdb-cluster. There is also a docker image that you can use.

it was working well, thank you