Actually I dont think continuous replication does not retry indefinitely (sadly), As far as I remember, both continuous and one off replication use the same failure mechanism, failed replication attempts will retry the request a specified amount of times (max_replication_retry_count) defaulted to 10, each retry after a failure will wait double the last retry wait time (up to a maximum of 10 minutes)
https://github.com/apache/couchdb/blob/master/src/couch_replicator/src/couch_replicator_manager.erl#L472 https://github.com/apache/couchdb/blob/master/src/couch_replicator/src/couch_replicator_manager.erl#L635 On 4 September 2013 18:17, Jens Alfke <[email protected]> wrote: > Let’s say CouchDB #1 has a replication to CouchDB #2. And let’s say that > connections to #2 are failing — maybe the server is down, maybe a router > failed, etc. > > If the replication is continuous, then I know that #1 will keep retrying > indefinitely, although I think the interval between attempts keeps > increasing. (Right?) > > What if the replication isn’t continuous? Will the replicator keep > retrying until it’s finished the replication (and then stop), or will it > stop at some point and give up with an error? > > —Jens
