Thanks you both for time spent and for good suggestions. Hope you'll be glad to know that this solution seems to work very well. :)
Best regards, Vincenzo On Tue, Jan 30, 2018 at 10:55 AM, Alessandro Benedetti <a.benede...@sease.io > wrote: > "At last, please let me ask another question, is it true that after every > commit, even if I had only updated one document, the SolrCloud cache is > invalidated (i.e. Solr must open a new searcher)? > Because this what the second clients does, updating a document at time and > commit. > In other words, how is good/bad having multiple hard commit in a short time > (few seconds)? " > > Your affirmation is correct when you open a new Searcher. > Caches are invalidated and possibly warmed up again (if you configured to > do > that). > This is valid for both hard and soft commits when you open a new searcher. > To have visibility you need to open a new searcher. > > Given that, I would definitely not recommend one commit ( even a soft > commit > which is lighter than an hard one, but still not for free) per document. > The overhead will be consistent, especially if you update few documents per > second. > I would go with an auto hard and soft commit on the updater client as well. > You can set up a timing between the commits which is compatible with the > maximum latency you can accept for updates to show up. > Solr gives support to Near Real Time search (through soft commits), and it > is definitely possible to tune it for "seconds" updates, > but I always recommend to start from the most acceptable latency in updates > and then reduce it if necessary. > > I will mention again a very valid blog from Erick, which explains in detail > the different type of commits : > > https://lucidworks.com/2013/08/23/understanding- > transaction-logs-softcommit-and-commit-in-sorlcloud/ > > > > ----- > --------------- > Alessandro Benedetti > Search Consultant, R&D Software Engineer, Director > Sease Ltd. - www.sease.io > -- > Sent from: http://lucene.472066.n3.nabble.com/Solr-User-f472068.html > -- Vincenzo D'Amore
