Mailing-List: contact user-help@couchdb.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@couchdb.apache.org
Received-SPF: neutral (nike.apache.org: local policy)
MIME-Version: 1.0
In-Reply-To: 
 <CAL+Y1nvFNmd4Lu2=1PXnZb-t6BZWNZv-P-KLs488eY+R1LDqPw@mail.gmail.com>
References: 
 <CAN5SjAC42fW=Q7au0MJeSCSfRJLYW3nw3qxO7Zmpr2XoSKVDMw@mail.gmail.com>
	<CAL+Y1nvFNmd4Lu2=1PXnZb-t6BZWNZv-P-KLs488eY+R1LDqPw@mail.gmail.com>
Date: Fri, 5 Oct 2012 18:03:25 +0200
Message-ID: 
 <CAL+Y1nv=cthDtzz95xy0d+=ZRRSGT5jYuoOYoFzsXFo5MUxqAg@mail.gmail.com>
Subject: Re: replication: lexical error: invalid char in json text
From: Dave Cottlehuber <dch@jsonified.com>
To: user@couchdb.apache.org
Content-Type: text/plain; charset=KOI8-R
Content-Transfer-Encoding: quoted-printable

On 5 October 2012 17:51, Dave Cottlehuber <dch@jsonified.com> wrote:
> On 5 October 2012 16:33, Michael Bykov <m.bykov@gmail.com> wrote:
>> Hi,
>>
>> Local couchdb works great, but replication does not work.

>> Replication worked perfectly several days ago. You can see the correct
>> (that was replicated) part of it here:
>> http://diglossa.ru:5984/_utils/database.html?greek
>>
>> Please advice how can I find invalid record in local DB?
>>
>> I got this in log:
>>
>>
>> =3DCRASH REPORT=3D=3D=3D=3D 5-Oct-2012::18:16:23 =3D=3D=3D
>>   crasher:
>>     initial call: couch_replicator:init/1
>>     pid: <0.720.0>
>>     registered_name: []
>>     exception exit: {worker_died,<0.727.0>,
>>                         {{nocatch,
>>                              {invalid_json,
>>                                  {{error,
>>                                       {1,
>>                                        "lexical error: invalid char in
>> json text.\n"}},
>>                                   <<"<html>\r\n<head><title>413
>> Request Entity Too Large</title></head>\r\n<body
>> bgcolor=3D\"white\">\r\n<center><h1>413 Request Entity Too
>> Large</h1></center>\r\n<hr><center>nginx/1.1.9</center>\r\n</body>\r\n</=
html>\r\n">>}}},
>>                          [{ejson,nif_decode,1,[{file,"ejson.erl"},{line,=
57}]},
>>                           {ejson,decode,1,[{file,"ejson.erl"},{line,38}]=
},
>>                           {couch_replicator_httpc,process_response,5,
>>                               [{file,"src/couch_replicator_httpc.erl"},
>>                                {line,88}]},
>>
>>
>>
>> --
>> =ED.
>>
>> http://diglossa.ru
>> xmpp://m.bykov@jabber.ru
>
> I'm guessing the underlying issue is replication with source being an
> older version of couchdb, or for some reason (e.g. different
> spidermonkey or JSON parsing library) is unable to complete parsing
> one document, causing the replication to die when only a partial doc
> is sent. It could be a lot more informative about where it's at
> however. If you're using filtered replication perhaps that may also be
> a mode of failure.
>
> Um there are a couple of ways to do this; the simplest is to sort &
> then diff the _all_docs list from both DBs, with whatever sensible
> tidyup is needed.
>
> Alternatively, I think in either source or destination log (when
> you're in debug mode) will show the last successful doc received or
> transferred.
>
> A+
> Dave

Actually after reading the HTML bit the issue is more likely with your
nginx config:

    413 Request Entity Too Large

I'd still diff _all_docs and then use single doc replication [1] as a
quick test, instead of needing to re-run the whole replication.

A+
Dave

[1]: http://wiki.apache.org/couchdb/Replication#Named_Document_Replication