Hi, I will try that approach. Deleting and force merging before adding the blocks. In my case, yes there are standalone docs (without any parents) and then there is blocks with parents and its children in the same index. Note however that docs in the blocks are unique in that the children, there is just one copy of the children. (Not sure if that was clear enough).
On 02/03/16, 9:40 PM, "Mikhail Khludnev" <mkhlud...@griddynamics.com> wrote: >Hello, > >It's really hard to find exact case, why it happens. There is a bruteforce >approach, sweep all deleted documents ie forcemerge until there is no >deleted docs. >Can it happen that standalone docs and parent blocks are mixed in the >index? > >On Wed, Mar 2, 2016 at 2:04 PM, Sathyakumar Seshachalam < >sathyakumar_seshacha...@trimble.com> wrote: > >> Am running in to this issue : >> https://issues.apache.org/jira/browse/SOLR-7606. But am not following >>all >> of the description there in that ticket. >> >> But what I am not able to understand is when does a parent/child >> orthogonality is broken. And what does a child document without a parent >> mean ? >> >> I have a set of documents that have been added to solr (via an import >>from >> DB), And then in another process I fetch or recreate SolrInputDocument >>from >> DB those documents for which the relation need to be in place. And >>before >> adding them to Solr, I make sure all the parent and child documents are >> deleted from Solr and then I add this block of documents in to solr. >> >> And now when I query I get an ArrayIndexOutOfBoundException exactly as >> specified in that JIRA issue. >> Any insights on how and what should be done will be greatly appreciated. >> >> >> >> > > >-- >Sincerely yours >Mikhail Khludnev >Principal Engineer, >Grid Dynamics > ><http://www.griddynamics.com> ><mkhlud...@griddynamics.com>
