Thank you both for your helpful replies! So, to summarize: * Running separate generic apache solr server is doable * Although ideal approach would be to use solr server 4.0 to take advantage of the cloud capability, there are compatibility problems with the index structure of dspace 3.0 that would need to be resolved first. So perhaps sometime soon this will become the best approach * Modifying dspace-stats/src/main/java/org/dspace/statistics/SolrLogger.java is a feasible approach to resolving for now and for making that feature more robust
A couple of notes and follow-up questions: Thanks, helix84, for the invitation to submit a proposed change to the Jira and to add some wiki documentation regarding redundancy. I hadn't meant for maintaining our dspace installation to become a full-time job, but I will submit the code change proposal to the Jira and promise to put the wiki reduncancy documetation on my to-do list once I get some other work done. João, I don't see where the "very bad performance" would happen if we were to use a separate solr server. Is your concern the delay incurred while spinning up the redundant solr server? Or am I missing something? All that said, I think probably I'll just do the code tweak for now, unless our office Linux guru talks me into the separate solr server. Running a single instance of a separate older solr server (to be compatible with our dspace 1.8.2) with failover wouldn't be such a chore for us, as we already do the same thing for our postgres database. And of course all dependent on any new info from João et al. Thank you both, again, for your responses. --Mike Reynolds On Thu, Jan 24, 2013 at 6:50 PM, João Melo <jm...@lyncode.com> wrote: > > Hi Michael, > > if losing statistical data isn't critical i think the first solution would be > enough. The second solution would work, as a tweak to achieve solr server > redundancy, however with very bad performance. It's possible to implement a > better redundant solution using the SolrReplication feature with a couple of > code & configuration changes, but i think that it wouldn't worth the effort. > > It would be better to focus the effort upgrading DSpace in order to use the > most recent Solr Cloud feature (Solr 4.1 - > http://wiki.apache.org/solr/SolrCloud). About the upgrade, from what i know, > Solr 3.5 indexes are readable from 4.1, however DSpace uses SolrJ Client v3.6 > which is incompatible with the newly Solr Cloud features, basically it will > only require code changes (and minor schema changes) - indexes will not > require upgrade. > > > > On 25 January 2013 01:37, helix84 <heli...@centrum.sk> wrote: >> >> Hi Michael, >> >> I'm glad you managed to get it working and that you reported back on your >> progress. Also, when you feel you have added something extra that others >> could use, you're welcome send the patch to our Jira. Howtos are also very >> welcome for our wiki! I don't think we have anything on redundancy yet. And >> the Solr page I wrote could use improvements in many places! >> >> Regarding the best solution for Solr availability, I would recommend you to >> run a separate Solr server if you can. There's nothing specific about the >> embedded Solr in DSpace - just copy over the statistics core (both conf and >> data), point DSpace(s) to it and you're good to go. The DSpace specific >> things are described by schema.xml, which lives in the core's conf >> directory. Just mind Solr versions, DSpace 3.0 contains Solr 3.5.0. >> >> Even if you don't have a machine to spare, you could always run 2 tomcat >> instances on one server to keep Solr safe in case DSpace crashes or locks up >> its Tomcat instance (usually because of memory exhaustion). If you want to >> go even further, Solr is designed so that it can be made redundant (see eg. >> [1] or [2]). >> >> [1] http://wiki.apache.org/solr/CollectionDistribution >> [2] http://wiki.apache.org/solr/SolrReplication >> >> >> Regards, >> ~~helix84 >> >> Compulsory reading: DSpace Mailing List Etiquette >> https://wiki.duraspace.org/display/DSPACE/Mailing+List+Etiquette >> >> >> ------------------------------------------------------------------------------ >> Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS, >> MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current >> with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft >> MVPs and experts. ON SALE this month only -- learn more at: >> http://p.sf.net/sfu/learnnow-d2d >> _______________________________________________ >> DSpace-tech mailing list >> DSpace-tech@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/dspace-tech >> List Etiquette: >> https://wiki.duraspace.org/display/DSPACE/Mailing+List+Etiquette > > > > > -- > Thanks, João Melo (My Portfolio) > DSpace Department > Lyncode: Official website ------------------------------------------------------------------------------ Master Visual Studio, SharePoint, SQL, ASP.NET, C# 2012, HTML5, CSS, MVC, Windows 8 Apps, JavaScript and much more. Keep your skills current with LearnDevNow - 3,200 step-by-step video tutorials by Microsoft MVPs and experts. ON SALE this month only -- learn more at: http://p.sf.net/sfu/learnnow-d2d _______________________________________________ DSpace-tech mailing list DSpace-tech@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dspace-tech List Etiquette: https://wiki.duraspace.org/display/DSPACE/Mailing+List+Etiquette
