If you use replication across multiple Solr indices, you can dedicate one just to indexing. For example, out of 3 total indices, 2 are used to serve queries and the third is for indexing. The indexing server is a master, and the query servers are slaves. In that scenario, the indexing server may be down for a period of time, but the query servers remain up. The data ends up synchronizing without query server downtime.
See the solrconfig.xml information here: https://vufind.org/wiki/vufind2:solr_replication
first of all I’d like to wish you all a Happy New Year!
But now to business:
We think about giving logged in users the opportunity to real time integrate metadata to the Solr index of our Vufind installation (1.x). In our tests we used a simple php form to fetch the data, we validated the content, generated the import file and sent it to the index.
Problem is that the index has to be rebuilt and the system is busy for over 60 seconds afterwards.
Is there a feasible way to do it without this time gap?
I know that normally the system’s not made for such actions but maybe it’s possible anyway?
Thanks in advance for your feedback!
GESIS - Leibniz Institute for the Social Sciences
Computational Social Science (CSS)
Team Knowledge Discovery
Jan Steinberg, M.A.
Unter Sachsenhausen 6-8, D-50667 Köln
Tel: + 49 (0) 221 / 47 694 – 215
Fax: + 49 (0) 221 / 47 694 – 560