couchdb-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Joan Touzet <woh...@apache.org>
Subject Re: _replicate vs. _replicator
Date Fri, 17 Jun 2016 20:38:29 GMT
Hi Ben, did anyone ever get back to you on this? I know that the
core developers have a LOT of reservations about the _replicator database,
primarily the fact that the backing store for a _replicator endpoint
probably shouldn't be a database itself (though it could conceivably present
a similar API to one).

Personally I still use the _replicate endpoint most of the time.

-Joan

----- Original Message -----
From: "Ben Keen" <ben.keen@gmail.com>
To: dev@couchdb.apache.org
Sent: Friday, June 3, 2016 10:11:30 AM
Subject: _replicate vs. _replicator

Hey!

Wonder if I could get some advice here. I’ve been working on refactoring
the replication feature in Fauxton to POST to the /_replicator database
rather than using /_replicate.

Having all replications (continuous/one-offs) logged in one place (the
_replicator database) leaves a nice paper trail of replication history.
[N.B. I’ve been speaking to Markus Fischboeck, who’s doing work on adding
some advanced replication features - we’re working in parallel].

I’ve been able to get replications working, but it requires passing both
basic headers in the POSTed JSON content, and the creds in the actual
endpoint URL, like so:

http://bob:bobspassword@mylocalsite.dev:8000/_replicator

It’s the latter that particularly worries me. I don’t believe this is
secure over http (correct?), and since Fauxton could be run anywhere, I
 wanted to know if I should stop heading down this road and stick with
_replicate.

Thanks!

Ben

Mime
View raw message