You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@couchdb.apache.org by "Randall Leeds (JIRA)" <ji...@apache.org> on 2011/01/26 23:36:44 UTC

[jira] Commented: (COUCHDB-1041) Send replication "conflict (are you replicating to yourself?)" error back

    [ https://issues.apache.org/jira/browse/COUCHDB-1041?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12987266#action_12987266 ] 

Randall Leeds commented on COUCHDB-1041:
----------------------------------------

I'm +1 on having this be a failure condition for replication and a 409 makes sense to me.
Here's at least one case where the error will not recover: save a checkpoint locally but it fails (time out, network issue, etc) remotely.
At the very least CouchDB should attempt to recover by reading the _rev from the remote checkpoint document so it doesn't get the same conflict again next time it checkpoints.

In most scenarios, except long-running replications (which going forward should be in the replicator db), this should maybe be failure. For transient network issues I would be fine with this case stopping the replication so long as it doesn't prohibit the replicator db from restarting it. I think this fits the duties of the replicator db and prevents the experience Benjamin reports.

> Send replication "conflict (are you replicating to yourself?)" error back
> -------------------------------------------------------------------------
>
>                 Key: COUCHDB-1041
>                 URL: https://issues.apache.org/jira/browse/COUCHDB-1041
>             Project: CouchDB
>          Issue Type: Improvement
>          Components: HTTP Interface
>    Affects Versions: 1.0.1, 1.1
>            Reporter: Benjamin Young
>            Priority: Minor
>
> I have three local CouchDB instances on various ports (differing versions). I accidently entered the port number for the instance I was replicating *to* in the *from* field, so the replication would have been to and from the same database. CouchDB doesn't currently send that error to the client, though it does output this in the logs: "checkpoint failure: conflict (are you replicating to yourself?)"
> It would be handy (at least for Futon) to have that information in either the response body or in the header. The log error message does include "conflict" in the wording, so perhaps a 409 Conflict error message is in order.
> Happy to provide a patch if 409 is the way to go.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.