couchdb-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Randall Leeds <randall.le...@gmail.com>
Subject Re: Replication -- explaining errors
Date Fri, 26 Mar 2010 20:20:17 GMT
On Fri, Mar 26, 2010 at 03:02, Germain Maurice
<germain.maurice@linkfluence.net> wrote:
> Hi all,
>
> Still with my problems with replication.
>
> I will write you a report on a crash of couchdb happened this night but now
> i launched again continuous replication hostA to hostB and i get this error
> on hostA :
>
> [Fri, 26 Mar 2010 09:55:01 GMT] [debug] [<0.2466.0>] retrying
> couch_rep_httpc post request in 16.0 seconds due to {error, req_timedout}
> [Fri, 26 Mar 2010 09:56:13 GMT] [debug] [<0.2466.0>] retrying
> couch_rep_httpc post request in 32.0 seconds due to {error, req_timedout}
> [Fri, 26 Mar 2010 09:57:42 GMT] [debug] [<0.2466.0>] retrying
> couch_rep_httpc post request in 64.0 seconds due to {error, req_timedout}
> [Fri, 26 Mar 2010 09:59:40 GMT] [debug] [<0.2466.0>] retrying
> couch_rep_httpc post request in 128.0 seconds due to {error, req_timedout}

This means that during replication, Couch has issued a post request
(most likely this means it's  /_ensure_full_commit during a
checkpoint) and that it timed out. The retry time backs off
exponential. Replication will crash after 10 retries.

There was a bug on 0.10.1 with replication crashing. I attempted to
fix some of the causes, but there are still some issues it seems. I
can reproduce this in production too, and I've had no luck tracking it
down yet. I'm going to re-open the 597 ticket and continue the
discussion there.

-Randall

Mime
View raw message