Question about Solr Cloud

[Cool Techi <co...@outlook.com>]

Hi,

I have just started working with Solr cloud and have a few questions related to the same,

1) In the start script we provide the the following, what's the purpose of providing this.
-Dbootstrap_confdir=./solr/collection1/conf Since we don't yet have a config in zookeeper, this parameter causes the local configuration directory ./solr/conf to be uploaded as the "myconf" config.  The name "myconf" is taken from the "collection.configName" param below.   -Dcollection.configName=myconf sets the config to use for the new collection. Omitting this param will cause the config name to default to "configuration1" 
2) When we make any changes into the config/schema do we need to copy it to all the shards running in the cloud manually?3) If we want to start with 10 shards on 2 machines, anticipating the future growth, do all these shards needs to run on separate jetty instances4) Is there any advantage of running solr on jetty then Tomcat?
Thanks,Ayush 

 		 	   		  

Re: Question about Solr Cloud

[Mark Miller <ma...@gmail.com>]

bq. fetch the configuration and store it locally.

New nodes don't fetch the configs and store them locally - configs are
loaded straight from zookeeper currently.

- Mark

Re: Question about Solr Cloud

[Upayavira <uv...@odoko.co.uk>]

You can force Solr to use the new configs by reloading a collection:

http://localhost:8983/solr/admin/collections?action=RELOAD&name=mycollection

This'll cause all shards (and replicas) in a collection to collect new
configs from ZooKeeper.

The main thing to note re Jetty, is that the Jetty included within Solr
is included for ease of demoing Solr, rather than for ease of
deployment. Whether you are going to deploy to Tomcat, JBoss or Jetty,
you would be best downloading a copy of the container, and installing
Solr within it (the one embedded doesn't have any startup scripts, nor
any maintenance interfaces, etc, all stuff that you'd expect from a
servlet container).

Upayavira



On Sat, Nov 17, 2012, at 03:04 PM, Erick Erickson wrote:
> 1> Well, it loads the local conf directory up to zookeeper so new nodes
> can
> fetch the configuration and store it locally.
> 2> No, you have to upload the configuration to ZK and (I think) restart
> the
> other servers. It's easy enough to test, just make your changes to the
> config, upload it, and look at the resulting configs to insure that the
> changes have been fetched.
> 3> No. You can run these shards in the same JVM as far as I know. This is
> sometimes called "microsharding" or "oversharding" and is a pretty common
> approach. Search the list I think theres been some discussion recently on
> this very topic.
> 4> Mostly the container you use is determined by which one you're
> comfortable with. Solr runs on Jetty, Tomcat, JBOSS and a host of others.
> It's just simplest to start OOB with Jetty.
> 
> Best,
> Erick
> 
> 
> On Sat, Nov 17, 2012 at 2:13 AM, Cool Techi <co...@outlook.com>
> wrote:
> 
> > Hi,
> >
> > I have just started working with Solr cloud and have a few questions
> > related to the same,
> >
> > 1) In the start script we provide the the following, what's the purpose of
> > providing this.
> > -Dbootstrap_confdir=./solr/collection1/conf Since we don't yet have a
> > config in zookeeper, this parameter causes the local configuration
> > directory ./solr/conf to be uploaded as the "myconf" config.  The name
> > "myconf" is taken from the "collection.configName" param below.
> > -Dcollection.configName=myconf sets the config to use for the new
> > collection. Omitting this param will cause the config name to default to
> > "configuration1"
> > 2) When we make any changes into the config/schema do we need to copy it
> > to all the shards running in the cloud manually?3) If we want to start with
> > 10 shards on 2 machines, anticipating the future growth, do all these
> > shards needs to run on separate jetty instances4) Is there any advantage of
> > running solr on jetty then Tomcat?
> > Thanks,Ayush
> >
> >

Re: Question about Solr Cloud

[Erick Erickson <er...@gmail.com>]

1> Well, it loads the local conf directory up to zookeeper so new nodes can
fetch the configuration and store it locally.
2> No, you have to upload the configuration to ZK and (I think) restart the
other servers. It's easy enough to test, just make your changes to the
config, upload it, and look at the resulting configs to insure that the
changes have been fetched.
3> No. You can run these shards in the same JVM as far as I know. This is
sometimes called "microsharding" or "oversharding" and is a pretty common
approach. Search the list I think theres been some discussion recently on
this very topic.
4> Mostly the container you use is determined by which one you're
comfortable with. Solr runs on Jetty, Tomcat, JBOSS and a host of others.
It's just simplest to start OOB with Jetty.

Best,
Erick


On Sat, Nov 17, 2012 at 2:13 AM, Cool Techi <co...@outlook.com> wrote:

> Hi,
>
> I have just started working with Solr cloud and have a few questions
> related to the same,
>
> 1) In the start script we provide the the following, what's the purpose of
> providing this.
> -Dbootstrap_confdir=./solr/collection1/conf Since we don't yet have a
> config in zookeeper, this parameter causes the local configuration
> directory ./solr/conf to be uploaded as the "myconf" config.  The name
> "myconf" is taken from the "collection.configName" param below.
> -Dcollection.configName=myconf sets the config to use for the new
> collection. Omitting this param will cause the config name to default to
> "configuration1"
> 2) When we make any changes into the config/schema do we need to copy it
> to all the shards running in the cloud manually?3) If we want to start with
> 10 shards on 2 machines, anticipating the future growth, do all these
> shards needs to run on separate jetty instances4) Is there any advantage of
> running solr on jetty then Tomcat?
> Thanks,Ayush
>
>