On 12/11/2014 9:19 AM, Michael Della Bitta wrote: > Only thing you have to worry about (in both the CUSS and the home grown > case) is a single bad document in a batch fails the whole batch. It's up > to you to fall back to writing them individually so the rest of the > batch makes it in.
With CUSS, your program will never know that the batch failed, so your code won't know that it must retry documents individually. All requests return with an apparent success even before the data is sent to Solr, and there's no way for exceptions thrown during the background indexing to be caught by user code. If your program must know whether your updates were indexed successfully by catching an exception when there's a problem, you'll need to write your own multi-threaded indexing application using an instance of HttpSolrServer. I filed an issue on this, and built an imperfect patch. The patch can only tell you that there was a problem during indexing, it doesn't know which document or even which batch had the problem. https://issues.apache.org/jira/browse/SOLR-3284 Thanks, Shawn