Hi, We have 5 core masters and 5 core slaves. The main core houses about 85,000 douments, so small, although the content of each document is quite large. The second core holds the same number of docs but far less - and different - data.
We reindex all cores every morning and the replication poll is 5 minutues. The main core takes 15 minutes to reindex (optimize). At some point, an incomplete index is picked up by the slave and our web site disappears until the optimize takes place. I know we could increase the poll to 30 minutes but that would be no guarantee. Thought we'd sove it by writing a script to get the indexversion, kick off reindexing and periodically check the current indexversion against the first - if the same, sleep for 2 minutes and then check again. Once they're different, do a fetchIndex from the slave. Works on all the cores except the main one. We get a different indexversion after two minutes, the slave gets populated with an almost empty index and the site is out! All the other cores exhibit the same indexversion. What have we misunderstood or got wrong? Regards, David Q -- View this message in context: http://lucene.472066.n3.nabble.com/indexVersion-returns-multiple-results-when-called-tp4036046.html Sent from the Solr - User mailing list archive at Nabble.com.
