On 1/7/2014 12:55 PM, jhittner wrote: > Hi, I'm working on setting up my first SolrCloud instance and have a three > questions I was hoping for some help with. I am running Solr 4.5.1. > > 1- I have 2 different schema's and 2 solrconfigs that I currently place in a > single conf folder that get uploaded to zookeeper automatically by tomcat > (bootstrap_confdir). I have been editing the core.properties files after > creating my collections to add the "schema=" and "config=" params to > instruct each core on what config file and schema to use. How can I > configure these two options when calling the API so I don't need to login to > the boxes and edit anything manually. The command I currently use looks > something like: > http://solrvip:8080/solr/admin/collections?action=CREATE&name=zipcollection-a&numShards=2&replicationFactor=2&maxShardsPerNode=4 > > 2- I store all my configs in a specific directory that I set with the tomcat > "bootstrap_confdir" option. After creating my collection using the API > call above, the collection runs perfect until the next restart. After > restarting, solr will not start back up until I create a "conf" directory in > each of the shards directories created by the API. The directory can be > empty, but needs to exist in order for Solr to run. Is this a bug or am I > doing something wrong? > > 3- As I stated in question 1, I put all my schema's and solrconfigs in a > single directory uploaded with the "bootstrap_confdir" directive. Is > there any way to organize this in sub-directories. I've tried several > ways, but can't seem to get it to work. for example: > bootstrap_confdir=/solr/conf > /solr/conf/collection-a/schema.xml > /solr/conf/collection-b/schema.xml
Let's take a step back and examine a more ideal way to do things. The bootstrap startup options are intended to be used once, and only to convert a non-cloud setup to a cloud setup. After that's done, they should be removed. In many cases, the only SolrCloud startup parameter you really need is zkHost: -DzkHost=zoo1.example.com:2181,zoo2.example.com:2181,zoo3.example.com:2181/solr In this example, there are three zookeeper hosts (the minimum required for a redundant setup), and Solr is using "/solr" as a chroot so the zookeeper root is clean. Note that you can put your zkHost value in your solr.xml file so that no special startup options are required at all. In my opinion, the bootstrap options shouldn't be used. Instead, you should use zkcli scripts to upload your config sets to zookeeper and then just create collections that reference the config set you want to use. If you need to change the config, you can change the local copy and then re-upload the config to zookeeper and do a RELOAD action on whichever collections are using that config. On the disk, a config set consists of a directory with solrconfig.xml, schema.xml, and any other config files referenced by those two xml files. When uploaded into zookeeper, all the files in that directory will be associated with a configName. Only the copy in zookeeper will actually be used. Here's information on the zkcli command-line script: https://cwiki.apache.org/confluence/display/solr/Command+Line+Utilities To create a collection once a config set is uploaded, use a URL like the following. For this example, the uploaded config is named "config1" and the new collection is named "test1". I have left out the maxShardsPerNode parameter, so this command would require that you have at least four Solr nodes in your cloud. http://server:port/solr/admin/collections?action=CREATE&name=test1&numShards=2&replicationFactor=2&collection.configName=config1 There should be no need to edit anything, and you will not need to tell Solr or any of the cores what file to use for solrconfig or schema. Thanks, Shawn
