Heya Kerr,

thanks for writing!

> On 13 Jan 2016, at 14:43, Kerr Rainey <[email protected]> wrote:
> 
> Hi All,
> 
> I’ve not looked at CouchDB for a long time and it’s great to see 2.0 getting 
> a bit of movement. Fantastic job everyone!

Thank you :)


> I’ve been messing about with CouchDB in docker and have a few questions.
> 
> I see that there is a dockerfile in the main apache/couchdb repo as well as 
> everything in the apache/couchdb-docker repo. Both have been updated 
> recently. Do they have different defined purposes? I can see the actual 
> differences, but I’m wondering what the plan is here.

In rough terms, the apache/couchdb/Dockerfile is to get someone, who wants to 
write a patch for CouchDB, up and running quickly without having to mess with 
installing the right Erlang version and whatnot. apache/couchdb-docker is more 
for production deployments.


> With regard to the apache/couchdb-docker repo, at a glance it looks identical 
> to the klaemo/docker-couchdb  "🐳Source of the official Apache CouchDB Docker 
> image” repo. I’m a little hazy on the github details here in terms of forks 
> and how these are linked. Clearly Clemens is doing the bulk (all?) of the 
> work here (great job, thanks), and the activity seems to happening “over 
> there”. I notice also that the request to docker-library/official-images for 
> the official CouchDB Docker images point to klaemo/docker-couchdb. This seems 
> a little confusing to me. Am I missing something? Planed, just not there yet?

This is mostly an infrastructure situation. klaemo/docker-couchdb was donated 
to the ASF by klaemo and is now tracked at apache/couchdb-docker. However, ASF 
infra is not yet 100% ready to just take over GitHub projects 1:1, and we’d 
love to move everything over including issues and all, and that’s still being 
worked on. So for the time being, we have both.


> Finally, obviously there are many scenarios for dockerizing couchdb, and it 
> looks like the current dev one is geared to testing out a little cluster, 
> which is probably what we need right now. For production I suspect that it 
> would rarely be a good idea to have more than one CouchDB server running in a 
> single container. Presumably we could explore some good patterns for 
> production deployments of couchdb in docker and have those images readily 
> available.

Definitely!


> I’m actually having a problem running klaemo/couchdb:2.0-dev image, failing 
> to start because of a connection error [Errno 111] while running cluster 
> setup (not the expected error when check nodes). I’ve not dug into it yet, 
> but please shout if it’s a known issue.

A few connection errors are expected (this is a cluster starting and a 
while/sleep loop checking if a node made it up yet), does it eventually finish, 
or do you get only connection errors after start up? Can you paste a log 
somewhere?

Best
Jan
-- 



> 
> Cheers
> Kerr
> 
> --
> Kerr Rainey
> 
> 
> 

Reply via email to