Isn't the designed workflow for something like this... 1) upload a very minimal configset 2) create your collection 3) upload your plugin jar 4) use the Schema API to create the new fieldtype that uses your plugin
? : Date: Wed, 7 Sep 2016 11:07:39 -0600 : From: John Bickerstaff <j...@johnbickerstaff.com> : Reply-To: solr-user@lucene.apache.org : To: solr-user@lucene.apache.org : Subject: Re: Can't create collection without plugin, : can't load plugin without collection : : I might be on the wrong track, but this statement made me think... : : I can force this to work by removing the custom reference, create the : collection, load the jar and then add the custom reference back in place. : This is fine as a manual one-time setup, but not feasible in a scripted : production deployment. : : If you're in a SolrCloud setup, it's only one step to change ALL the : configs, right? Just upload "improved" configs to zookeeper from one box, : restart all boxes. The improved configs could already reside on one (or : all) of your SolrCloud servers (vms or physical machines - or be on a : shared network drive) : : Unless I'm missing something (and I could be) it's fine for the jar to be : in the appropriate folder on each box as long as it's NOT referenced in the : config file... So that could be part of your build for every machine (or : you could use a script to deploy it) : : Just trying to help think around the corners of the box - feel free to : ignore if I've missed something obvious. : : : On Wed, Sep 7, 2016 at 10:44 AM, Erick Erickson <erickerick...@gmail.com> : wrote: : : > Well, it was worth a try ;).... : > : > On Wed, Sep 7, 2016 at 9:42 AM, John Bickerstaff : > <j...@johnbickerstaff.com> wrote: : > > Not sure if it's helpful, but I did solve a similar problem by scripting : > > the placement of the plugin .jar file into the appropriate directories as : > > part of the build process for my VMs : > > : > > I get that doing it through the BlobStore would be preferred, probably... : > > But I couldn't get it to work and decided to go with "ye olde school" : > bash : > > scripting which I knew I could depend on. : > > : > > (Caveat - I have a fairly simple environment, so your mileage may vary) : > > : > > On Wed, Sep 7, 2016 at 10:38 AM, jimtronic <jimtro...@gmail.com> wrote: : > > : > >> Sadly, that didn't work. : > >> : > >> Without a core to hit, the /[COLLECTION]/config returns a 404 error. : > >> : > >> The best bet at this point may be for me may be one of the following: : > >> : > >> 1. Programmatically modify configoverlay.json file to add the runtime : > libs : > >> when I upload the config. : > >> : > >> or : > >> : > >> 2. Patch solr so that schema.xml loads custom classes directly from the : > >> BlobStore like solrconfig.xml does. : > >> : > >> or : > >> : > >> 3. Patch solr so that you can specify configSets instead of a collection : > >> when associating a runtimeLib. : > >> : > >> : > >> : > >> : > >> : > >> : > >> -- : > >> View this message in context: http://lucene.472066.n3. : > >> nabble.com/Can-t-create-collection-without-plugin-can- : > >> t-load-plugin-without-collection-tp4294865p4295028.html : > >> Sent from the Solr - User mailing list archive at Nabble.com. : > >> : > : -Hoss http://www.lucidworks.com/