DSPACE HIGH AVAILABILITY

[Gina Alegre Milla]

Dear team.

In order to have high availability applied to my DSpace repository (7), I have set up a DSpace master and replica, the database is synchronized and assetstore is being shared to the replica via the NFS file system. Initially everything was running normal, the data was being replicated but when creating the collections and publications, the metadata of these is not shown.

Internally this information is stored in the database as well as in the assetstore, but in the browser I can only visualize the communities. Indeed in my master everything is working well, the problem is in the replica.

Please if you could comment if I am configuring the replication correctly or if I am ignoring something. Also if it is possible to use another method to configure DSpace in high availability.

thanks a lot!

[Mark H. Wood]

You haven't mentioned Solr. Is it working properly with the replica
DSpace instance?

Do you find anything of interest in DSpace's logs on the replica host?

--
Mark H. Wood
Lead Technology Analyst

University Library
Indiana University - Purdue University Indianapolis
755 W. Michigan Street
Indianapolis, IN 46202
317-274-0749
www.ulib.iupui.edu

[Gina Alegre Milla]

Dear, Mark

Solr is running correctly, what I was missing was to index the synchronized data.

After running /dspace/bin/dspace index-discovery  I could visualize the content of my repository (communities, collections and publications). Every time the assetstore is synchronized I have to run the indexing. I was planning to create a cron job to perform this process.

If there is another method to configure Dspace in high availability, please let me know your comments.

Thank you very much

[Jumail Mundekkat]

I have pulled off DSpace high availability setup on 6.3 by using a high availability setup of both Postgres and Solr,  and using an NFS share for the assetstore folder. As a read-only database this worked fine but the issue arose when you wanted to authenticate a user. A clustering solution is required for the session objects, I suspect it may be possible to do at a Tomcat level but I have not attempted it.

If you are needing to run index-discovery constantly, I would take a look at your clustering solution for Solr as it may not be correctly synchronising data between nodes.