[Apiman-user] Elasticsearch configuration

[Aikeaguinea]

I'll keep you posted. Using Amazon's Elasticsearch service presents
issues of its own, since it can't be secured using VPC security groups.
I found a very lightweight JS proxy for signing requests to the service,
which allows you to use IAM users and roles to restrict access to the
service -- this may also be useful knowledge for anyone who gets into
this situation.

The proxy is on Gist at
https://gist.github.com/nakedible-p/ad95dfb1c16e75af1ad5 with a
description at https://forums.aws.amazon.com/thread.jspa?threadID=218214
.

On Thu, Feb 25, 2016, at 07:50 AM, Eric Wittmann wrote:
> No problem.  If you end up with a production configuration strategy for 
> elasticsearch+apiman we'd love to hear about it!  Perhaps a guest blog 
> post?  :)
> 
> Just a thought...
> 
> -Eric
> 
> On 2/24/2016 1:52 PM, Aikeaguinea wrote:
> > Thanks!
> >
> > On Wed, Feb 24, 2016, at 10:44 AM, Eric Wittmann wrote:
> >> Apiman will automatically create the appropriate ES indexes if they do
> >> not exist.  However, it does this in a very naive way:
> >>
> >> https://github.com/apiman/apiman/blob/master/gateway/engine/es/src/main/java/io/apiman/gateway/engine/es/ESClientFactory.java#L220-L233
> >>
> >> That said, we aren't really elasticsearch experts, so I'm not too
> >> confident offering advice on how to scale/cluster a production instance
> >> of Elasticsearch (note that we are working on alternatives).
> >>
> >> My suggestion is probably to create the indexes manually, and shard them
> >> however you think is appropriate.  Apiman will then simply use the
> >> indexes you create rather than create them for you.  You can use the ES
> >> Mappings files when configuring the indexes:
> >>
> >> https://github.com/apiman/apiman/tree/master/gateway/engine/es/src/main/resources/io/apiman/gateway/engine/es
> >>
> >> -Eric
> >>
> >> On 2/23/2016 5:07 PM, Aikeaguinea wrote:
> >>> I'm moving toward a production deployment of apiman on AWS. At least for
> >>> now we're using Amazon's Elasticsearch service and have been able to
> >>> configure and use it.
> >>>
> >>> For production, we need to pay attention to some of the finer details of
> >>> clustering, etc. It looks like sharding is mainly controlled when an
> >>> index is created. How does apiman manage creating replica shards? Does
> >>> this need to be configured somehow?
> >>>
> >
> >


-- 
  Aikeaguinea
  aikeaguinea at xsmail.com

-- 
http://www.fastmail.com - A fast, anti-spam email service.