On 3/25/2016 2:04 AM, fabigol wrote: > what i want to do and to create the differents links between the entities > which i'm going to index. Therefore, i have a root entity and girls entities > like showing xml File. > > But, my main problem is the number of documents. In facr, when i want to > index 3 months of data i have no problem(5 millions), if i want to index 6 > months of data (10 millions) the indexation is not finished after 18 hours.
I'm still operating with very little information here. Best guess with the information I have: Your heap is too small and when the index size gets big enough, Solr is operating in an extremely low memory environment. Because you have caching on some of your DIH entities, which eats up heap memory quickly, this seems even more likely. If this is what's happening, Java is doing EXTREMELY frequent full garbage collections so there's enough memory for Solr to run. You might even be seeing occasional OutOfMemory exceptions in your log, which means that Java is actually running completely out of memory, and Solr's behavior becomes undefined. As I said, I'm guessing. I could be completely wrong, but without a lot more information, there's no way for me to know. If you're running Solr 5.x, you should have a garbage collection log. Analyzing that with a tool like gclogviewer might show you any GC problems. Thanks, Shawn
