Actually this is definitely not a ram issue. I have visualVM connected and MAX Ram available for the JavaVM is ~7GB, but the system is only using ~5.5GB, with a MAX so far of 6.5GB consumed.
I think..well I'm guessing the system hit a merge threshold, but I can't tell for sure..I have seen the index size grow rapidly today (much more then normal, in the last 3 hours the index size has increased by about 50%). >From various posts I see that during the 'optimize' (which I have not called), or the perhaps the merging of segments it is normal for the disk space requirements to temporarily increase by 2x to 3x. As such my only assumption is that it must be conducing a merge. Note: since I restarted the solr server, I have only 1 client thread pushing data in (it already transmitted the data.(~2mb)). (and it has been held up for about 4 hours now..I believe its stuck waiting for the merge thread to complete). Is there a better way to handle merging? or at least predicting when it will occur? (I'm essentially using the defaults MergeFactor:10, ramBuffer 32MB). I'm totally new to solr/lucune/indexing in generaly so I'm so what clueless on all this.. It should be noted we have 'millions of documents' all which are generally < 4k bytes. -- View this message in context: http://lucene.472066.n3.nabble.com/Solr-Hanging-all-of-sudden-with-update-csv-tp2652903p2653423.html Sent from the Solr - User mailing list archive at Nabble.com.
