CouchDB 1.6 Replication from two sources

i have a central CouchDB Instance which holds one database with my ddoc's and several databases for my data. The reason to put the ddoc's in a separate database resides in the fact that i use the central CouchDB Instance only as a distributer for other CouchDB Instances. Every database which is holding data uses the same ddoc's.

Central Couch: db_customer1 db_customer2 db_ddoc's

CouchDB customer1: db_customer1 + db_ddoc's

CouchDB customer2: db_customer2 + db_ddoc's

Testing

1.) To test the use case i created two databases in my central CouchDB Instance and inserted the appropriate data.

2.) After that i created a second CouchDB Instance and a database.

3.) The next step was to replicate the database with the data to the second CouchDB Instance. That works like a charm.

4.) The last step is now to replicate all the ddoc's. But this fails.

replication document:

 {
   "_id": "3cd50136342edc1cd9660e29f60148f4",
   "_rev": "3-3c058e1f8a2bee8b896719b728b5119b",
   "source": "design",
   "target": "http://robert:secret@172.17.0.1:5983/kunde_a",
   "owner": "robert",
   "_replication_state": "completed",
   "_replication_state_time": "2016-05-18T10:04:34+02:00",
   "_replication_id": "74498297514f19611af3f198e36beb7c",
   "_replication_stats": {
       "revisions_checked": 0,
       "missing_revisions_found": 0,
       "docs_read": 0,
       "docs_written": 0,
       "doc_write_failures": 0,
       "checkpointed_source_seq": 13
   }
}

CouchDB log from the central CouchDB Instance

Wed, 18 May 2016 08:09:42 GMT] [info] [<0.3840.1>] 127.0.0.1 - - POST /_replicator 201
[Wed, 18 May 2016 08:09:42 GMT] [info] [<0.135.0>] Attempting to start replication `74498297514f19611af3f198e36beb7c` (document `3cd50136342edc1cd9660e29f6014f12`).
[Wed, 18 May 2016 08:09:42 GMT] [info] [<0.4327.1>] Replication `"74498297514f19611af3f198e36beb7c"` is using:
        4 worker processes
        a worker batch size of 500
        20 HTTP connections
        a connection timeout of 30000 milliseconds
        10 retries per request
        socket options are: [{keepalive,true},{nodelay,false}]
        source start sequence 13
[Wed, 18 May 2016 08:09:42 GMT] [info] [<0.4327.1>] Document `3cd50136342edc1cd9660e29f6014f12` triggered replication `74498297514f19611af3f198e36beb7c`
[Wed, 18 May 2016 08:09:42 GMT] [info] [<0.4326.1>] starting new replication `74498297514f19611af3f198e36beb7c` at <0.4327.1> (`design` -> `http://robert:*****@172.17.0.1:5983/kunde_a/`)
[Wed, 18 May 2016 08:09:42 GMT] [info] [<0.4327.1>] Replication `74498297514f19611af3f198e36beb7c` finished (triggered by document `3cd50136342edc1cd9660e29f6014f12`)

CouchDB log from the second CouchDB Instance

172.17.0.1 - - HEAD /kunde_a/ 200
[Wed, 18 May 2016 08:04:34 GMT] [info] [<0.5521.0>] 172.17.0.1 - - GET /kunde_a/ 200
[Wed, 18 May 2016 08:04:34 GMT] [info] [<0.5521.0>] 172.17.0.1 - - GET /kunde_a/_local/74498297514f19611af3f198e36beb7c 200

If i first insert the ddoc's and let the replicator create a new database the process completes, but if i then replicate my data database the ddoc's vanish and i have the same problem as before.

If both, ddocs and data, are in one database the replication works flawless.

Any help is welcome :)