couchdb-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Filipe David Manana <fdman...@apache.org>
Subject Re: Kill/exit of thread during pull replication
Date Tue, 23 Aug 2011 20:34:02 GMT
Martin, it's good to know it's not happening always and you're able to
do your replications.

My guess of what happened:

Couch started and it attempted to start (restart) replications
persisted in the _replicator database.
These replications didn't get started in about 20 seconds (maybe the
target was not up yet? or very slow/congestioned network), so the
supervisor stopped the attempt and waited for a while before retrying
- I'm supposing this from the following entry in the log:

Tue, 23 Aug 2011 16:39:24 GMT] [error] [<0.110.0>] Error in
replication `82a20dd08c23464fd396eae220bc054a` (triggered by document
`36ac5e2869cfd8e485106c9e710013de`): timeout
Restarting replication in 5 seconds.

The stack trace happens because a process was a killed, by the
supervisor, after those 20 seconds.
If this happens very often, specially not at server startup, please
open an issue at:  https://issues.apache.org/jira/browse/COUCHDB

On Tue, Aug 23, 2011 at 11:29 AM, Martin Hewitt <martin@thenoi.se> wrote:
> Hi Felipe,
>
> Subsequent replication operations have been created and run to completion without the
error. These first few replications, the ones that triggered errors, were the first ever actions
on a freshly-built, completely empty CouchDB instance.
>
> I haven't gone through it line-by-line, but yes, it seems the replication completed without
much more error.
>
> There are more errors, but all very much within the first 3 minutes of the server being
started and the replications being started on it.
>
> I'm not sure which errors are recurrent, but I can go through each stack and see which
are repeated.
>
> I've POSTed further pull replications to the _replicator database, and they've run and
completed without error.
>
> Martin
>
> On 23 Aug 2011, at 19:11, Filipe David Manana wrote:
>
>> Martin,
>>
>> can I ask you to try to POST to /_replicate (instead of the
>> _replicator database) and see if it gets better?
>>
>> Looking at your document, it seems the replication completed, despite
>> the stack trace. Is it true?
>>
>> thanks
>>
>> On Tue, Aug 23, 2011 at 10:50 AM, Martin Hewitt <martin@thenoi.se> wrote:
>>> Hi Filipe,
>>>
>>> Sorry, I'm no good at deciphering ERLang stack traces, what bits am I missing?
I could post my entire log file, but I think that'd be overkill...
>>>
>>> I've updated the same Gist (https://gist.github.com/1165822) with as full a trace
as I can decipher from the first instance of the error.
>>>
>>> I started the replication by POSTing to _replicator, the document is here: https://gist.github.com/1165982
>>>
>>> Martin
>>>
>>> On 23 Aug 2011, at 18:40, Filipe David Manana wrote:
>>>
>>>> Martin,
>>>>
>>>> can you paste the full stack trace and how have you triggered the
>>>> replication (object/document)?
>>>>
>>>> On Tue, Aug 23, 2011 at 9:49 AM, Martin Hewitt <martin@thenoi.se> wrote:
>>>>> Hi all,
>>>>>
>>>>> Having finally gotten trunk compiled and running, I began a pull replication
from our production server to test some bits locally.
>>>>>
>>>>> The following exception keeps cropping up: https://gist.github.com/1165822
>>>>>
>>>>> I'm no good with erlang stack traces, so I can't tell if this is an error
with my setup or with Couch.
>>>>>
>>>>> There are no errors on the server being pulled from.
>>>>>
>>>>> Martin
>>>>
>>>>
>>>>
>>>> --
>>>> Filipe David Manana,
>>>> fdmanana@gmail.com, fdmanana@apache.org
>>>>
>>>> "Reasonable men adapt themselves to the world.
>>>>  Unreasonable men adapt the world to themselves.
>>>>  That's why all progress depends on unreasonable men."
>>>
>>>
>>
>>
>>
>> --
>> Filipe David Manana,
>> fdmanana@gmail.com, fdmanana@apache.org
>>
>> "Reasonable men adapt themselves to the world.
>>  Unreasonable men adapt the world to themselves.
>>  That's why all progress depends on unreasonable men."
>
>



-- 
Filipe David Manana,
fdmanana@gmail.com, fdmanana@apache.org

"Reasonable men adapt themselves to the world.
 Unreasonable men adapt the world to themselves.
 That's why all progress depends on unreasonable men."

Mime
View raw message