With SOLR-5115 there's support for forcing ZkResourceLoader to use
SolrResourceLoader using file:/// prefix in your schema. This forces Solr to
load files from the FS as usual.
https://issues.apache.org/jira/browse/SOLR-5115
-----Original message-----
> From:Markus Jelsma <markus.jel...@openindex.io>
> Sent: Friday 2nd August 2013 15:27
> To: solr-user@lucene.apache.org
> Subject: RE: Large config files in SolrCloud
>
> Ok, i managed to load a config file from the node's sharedLib directory by
> pointing it to `../../lib/file`. This works fine in normal mode, in cloud
> mode all it does is attempting to find it in Zookeeper by trying
> /configs/COLLECTION_NAME/path. Properties such as ${solr.home} are not
> recognized in the schema, it seems.
>
> Any idea on how to force Solr in cloud mode to grab the file from the local
> FS?
>
> Thanks,
> Markus
>
> -----Original message-----
> > From:Markus Jelsma <markus.jel...@openindex.io>
> > Sent: Friday 2nd August 2013 14:34
> > To: solr-user@lucene.apache.org
> > Subject: RE: Large config files in SolrCloud
> >
> > Yes, all the usual config files are well under 1MB and work as expected.
> > This file is under 2MB and the limit i set is 5MB. Setting jute.maxbuffer
> > (all lowercase) did work during a test a long time ago but we'd like to put
> > the new features in production and we're stuck at this trivial issue :)
> >
> > Thanks
> >
> > -----Original message-----
> > > From:Erick Erickson <erickerick...@gmail.com>
> > > Sent: Friday 2nd August 2013 14:28
> > > To: solr-user@lucene.apache.org
> > > Subject: Re: Large config files in SolrCloud
> > >
> > > Hmmm, does it work with smaller config files? There's been a limit of 1M
> > > for ZK files, and I'm wondering if your setup would work with, say, 2M
> > > configs as a check that it's "something else" rather than just the 1M
> > > limit.
> > >
> > > FWIW,
> > > Erick
> > >
> > >
> > > On Fri, Aug 2, 2013 at 8:18 AM, Markus Jelsma
> > > <markus.jel...@openindex.io>wrote:
> > >
> > > > Hi,
> > > >
> > > > I have a few very large configuration files but it doens't work in cloud
> > > > mode due to the KeeperException$ConnectionLossException. All 10 Solr
> > > > nodes
> > > > run trunk and have jute.maxbuffer set to 5242880 (5MB). I can confirm
> > > > it is
> > > > set properly by looking at the args in the Solr GUI. All Zookeepers have
> > > > exactly the same key/value set, i can confirm this by looking with ps at
> > > > the process, it is really there, the first parameter.
> > > >
> > > > But it doesn't work!
> > > >
> > > > I have had it working once but it doesn't like me anymore. Putting the
> > > > config files in the node's sharedLib or the core's lib directory doesn't
> > > > work either, the files are clearly loaded according to the logs but the
> > > > TokenFilters cannot access them and complain about a config file not
> > > > being
> > > > found.
> > > >
> > > > I'm out of ideas, any to share?
> > > >
> > > > Thanks,
> > > > Markus
> > > >
> > >
> >
>
