lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Noble Paul <noble.p...@gmail.com>
Subject Re: Solr 5 Jetty going away - which JIRA?
Date Thu, 17 Apr 2014 14:29:20 GMT
Can we have the choice of jetty connector or netty connector to the user?
The user should be able to provide appropriate connector properties for
each connector.


On Thu, Apr 17, 2014 at 2:31 PM, Uwe Schindler <uwe@thetaphi.de> wrote:

> Hi,
>
> SOLR-5091 is going in that direction, but that is not the issue about
> Jetty.
>
> In fact the main thing is: We don't want to go away just from Jetty, the
> idea is to remove the whole "web application" (including the WAR file,
> which is already gone in trunk). So we won't use any servlet api stuff,
> web.xml,... anymore. The idea is to go more in the direction of
> ElasticSearch that uses Netty as library for the HTTP access.
>
> The main reason is: Solr is not a web application that is sitting in the
> infrastructure of some application server we have no control of, it is more
> a separate Java application, running in its own VM and providing a HTTP
> connector to the outside. Like a postgres sql server or mysql. You would
> also never run such servers inside a application server. Possibly, there
> could also be a binary only, non-http connector (like ElasticSearch port
> 9300). The latter is not easy to implement if the whole thing is running as
> a child of a webapp container. Also the embedded Zookeeper is also
> something you should not do inside a webapp container. With a separate
> application, Solr would have full control on setting up the zookeeper and
> connectors to the outside. In fact we could also provide auto-discovery by
> sending broadcasts.
>
> Theoretically, instead of using Netty, we could still use "embedded Jetty"
> to provide the HTTP connector, but Solr would *not* be a web application
> nor we would implement one hidden inside. It would only run an embedded
> Jetty as one of the connectors to the outside. Starting Solr would go via a
> main() method somewhere (like ES's Bootstrap class) and some shell scripts
> that setup good defaults for garbage collectors and memory settings.
>
> But you are right, there is no issue at the moment, because it is not yet
> decided how the infrastructure should look like.
>
> Uwe
>
> -----
> Uwe Schindler
> H.-H.-Meier-Allee 63, D-28213 Bremen
> http://www.thetaphi.de
> eMail: uwe@thetaphi.de
>
>
> > -----Original Message-----
> > From: Alexandre Rafalovitch [mailto:arafalov@gmail.com]
> > Sent: Thursday, April 17, 2014 10:39 AM
> > To: dev@lucene.apache.org
> > Subject: Solr 5 Jetty going away - which JIRA?
> >
> > I've read a couple of times now that Jetty/Tomcat/Container is going
> away in
> > Solr 5.
> >
> > Can somebody tell me where to check the detailed discussion about that.
> Is
> > that a mailing list thread? A JIRA? Something else?
> >
> > I found SOLR-5091, but it does not feel quite right.
> >
> > Regards,
> >    Alex.
> > Personal website: http://www.outerthoughts.com/ Current project:
> > http://www.solr-start.com/ - Accelerating your Solr proficiency
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org For additional
> > commands, e-mail: dev-help@lucene.apache.org
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
> For additional commands, e-mail: dev-help@lucene.apache.org
>
>


-- 
-----------------------------------------------------
Noble Paul

Mime
View raw message