I believe it's because the default is for SolrCloud where you will have replicas of the info for recovery and so it's just overkill. The tran log is still used for other things (realtime get, peer sync - it may play a part in other features).
We should probably doc this well so that single node users that want the sync know to turn it on. Syncing so often is so expensive, you certainly would not want to do it in a cloud setup. bq. I guess, if the JVM crashes, you can recover the indexed documents... Yes, though I don't think that is a primary reason for anything. This is not my area though, so FWIW. On Fri, Jul 13, 2012 at 9:48 AM, Michael McCandless < [email protected]> wrote: > There was a thread a couple months back: > > http://markmail.org/thread/7w3z2lsgb7fpfewb > > That didn't reach closure, so I want to re-raise / answer it: > > It seems like SyncLevel defaults to FLUSH not SYNC in Solr's > TransactionLog? Is that still true in Lucene 4.0 alpha? > > Does this really mean on OS crash / power loss, documents are lost? > If so, why even write to the transaction log? > > (I guess, if the JVM crashes, you can recover the indexed documents...). > > Mike McCandless > > http://blog.mikemccandless.com > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > > -- - Mark http://www.lucidimagination.com
