Couch db replication not working - nano module unable to create replication job in _replicator database

[shantanunirale]

Couch db replication not working - nano module unable to create replication job in _replicator database.

Previously, in nano modules v4.1.1, it was working but for current version its giving positive response but it seems no replication job created.

Is this a new release change for nano module?

For V4.1.1 replicateDb function in nano has return something like "return relax({db: "_replicator", body: opts, method: "POST"}, callback);" But for the current version its showing me "return relax({db: '_replicate', body: opts, method: 'POST'}, callback);" And I am sure this change causes the replication job creation failure.

[jo]

Looking at the code of v4.1.1 (and also v4.0.0) Nano has even then used the _replicate endpoint, not the _replicator database:

    return relax({db: "_replicate", body: opts, method: "POST"}, callback);

fyi the _replicator database was introduced in CouchDB v1.1.0 and allows persistent replications, which are kept even after a CouchDB restart.

Does the replication show up under _active_tasks?

[shantanunirale]

Actually, if I try to use the nano module with this _replicate end point, The replication jobs are not created and even, it is not available in _active_tasks. The response of the replicator function is also even. Same as present in documentation { "ok": true, "_local_id": "0a81b645497e6270611ec3419767a584+continuous+create_target" }

But if I change the end point to _replicator. It is working as expected. Replication job entry got created under _replicator database. I am able to see the same replication job in _active_task also.

Could you please help me with figuring out whats going wrong?

[jo]

I have checked again it in current nano v6.1.5 with the following code:

var nano = require('nano')('http://localhost:5984')

nano.db.replicate('asd', 'efg', { create_target: true }, function(error, response) {
  console.log(error, response)
})

The response was:

{ ok: true,
  session_id: '89a68ff7c703e2cbda76d67108a8326e',
  source_last_seq: 4,
  replication_id_version: 3,
  history: 
   [ { session_id: '89a68ff7c703e2cbda76d67108a8326e',
       start_time: 'Tue, 01 Sep 2015 08:15:45 GMT',
       end_time: 'Tue, 01 Sep 2015 08:15:46 GMT',
       start_last_seq: 0,
       end_last_seq: 4,
       recorded_seq: 4,
       missing_checked: 2,
       missing_found: 2,
       docs_read: 2,
       docs_written: 2,
       doc_write_failures: 0 } ] }

indicating a successful replication.

[jo]

I have also issued a continuous replication:

nano.db.replicate('asd', 'efg', { continuous: true }, function(error, response) {
  console.log(error, response)
})

with the response:

{ ok: true,
  _local_id: '2971ed4ac2a08832adb79f26f051c8fa+continuous' }

I than checked _active_tasks:

curl http://localhost:5984/_active_tasks | json_pp

and it shows that the continuous replication is running:

[
   {
      "checkpoint_interval" : 5000,
      "checkpointed_source_seq" : 4,
      "doc_id" : null,
      "revisions_checked" : 0,
      "started_on" : 1441095695,
      "continuous" : true,
      "target" : "efg",
      "doc_write_failures" : 0,
      "source_seq" : 4,
      "progress" : 100,
      "missing_revisions_found" : 0,
      "type" : "replication",
      "pid" : "<0.1630.0>",
      "updated_on" : 1441095740,
      "docs_read" : 0,
      "replication_id" : "2971ed4ac2a08832adb79f26f051c8fa+continuous",
      "docs_written" : 0,
      "source" : "asd"
   }
]

So everything works as expected :)

[mukundnc]

Hi Jo,

The issue is not with the instance. Here you are trying to create a replication job on your local couchdb. Can you please try the same on http://www.cloudant.com, the problem here is we are unable to find the replication job on the cloudant server.

The major problem what we are facing here is:

1. When ever any maintenance happens on cloudant no replication happens across our DBs.
2. We cannot validate the replication jobs as the documents are not created on the cloudant server even though the module created a replication job on the server.

Thanks Shawn

[jo]

I'm not sure that I fully understand the problem. Do you suggest to drop the _replicate endpoint in favor of the _replicator database?

[shantanunirale]

As mentioned, we are facing issue with the replication job which is not created after replication. And I think which should be created for getting the entries in _active_tasks . Could you please try out on cloudant if you are facing the same issue?

[jo]

But what does that have yo do with nano?

[carlosduclos]

I'm implementing proper replication to the "_replicator" database. Follow on: https://github.com/carlosduclos/nano/tree/replicator Should be done today EOD or tomorrow.

[carlosduclos]

Working on this on #349.

[carlosduclos]

This issue has been solved by implementing replication using the _replicator database. Before that nano was only using _replicate and not _replicator. The difference in behavior explained by @shantanunirale is expected, both databases have very different behavior. When using _replicate, replication is started at once and the request is not ready until replication has run. When using _replicator, a new replication job is scheduled and the reply is the id of that job. Please look at apache/couchdb-nano to get the latest updates.

[SK-CSE]

@jo hello there, i am trying to replicate a document in _users db from one instance to another, but it is not returning any response, but to replicate within a instance it is working for some random db. please find the below code i am trying to execute.. nano.db.replicate('_users', 'http://admin:password@otherhost.com:5984/_users', { create_target:true, doc_ids:["org.couchdb.user:sk5"] }, function(err, result) { if (!err){ console.log(result) }else{ console.log(err2) } });