We just had a discussion in another thread about deleting and re-creating _replicator documents. Try adding a random or changing value to your document. Does that help?
I described detailed techniques for handling replicator documents in my earlier post. On Wed, Sep 25, 2013 at 6:40 PM, [email protected] <[email protected]> wrote: > Am 25.09.2013 12:41, schrieb Dave Cottlehuber: > >> Hey Marten, Sorry to hear things didn't work out. Some more details might >> help. How was your replication set up? What sort of replication, esp was it >> continuous? > > it was a continous replication with a filter function. the filter function > itself was ok > >> and what does "restart replication" mean? > > well, we stored stuff into the database, but that stuff was not replicated > to the other two databases - therefore we deleted the documents in the > replicator database (they were showing no error state) and stored it again - > but no task was assigned to these new documents. > >> Was it stored in the replicator db or not? > > yes > >> Any relevant errors in your system logs around that time? >> >> Its unusual for a restart of couchdb not to fix stuff, and when the >> network comes up couchdb should restart any replications saved in the >> replicator db automatically, assuming its not exceeded the (configurable) >> maximum restarts. >> >
