You're right, thanks!
2011/6/20 Erick Erickson <erickerick...@gmail.com>: > Hmmm, be a little careful here with terminology. > Shards may be unnecessary if you can put your whole index > on a single searcher. It's preferable to simply have each > slave hold a complete copy of the index, no sharding necessary. > > Best > Erick > > On Mon, Jun 20, 2011 at 9:45 AM, Mark Schoy <hei...@gmx.de> wrote: >> Thanks for your answer Erick. >> >> So the easiest way will be to set up 2 shard cluster with shard replicas ;) >> >> 2011/6/20 Erick Erickson <erickerick...@gmail.com>: >>> No, there's nothing built into Solr to automatically promote a slave >>> to a master. >>> >>> You have several choices here. One is to build a new master and >>> reindex from scratch. >>> >>> Another is to configure your slave as a new master and then >>> bring up a new machine and have it replicate. Now make that new machine >>> your master (you'll have to re-configure both). >>> >>> The fun part is continuing to serve requests while all this is going >>> on. It's easier >>> if you have more than one slave so you can move things around while the >>> remaining slave is reconfigured (or whatever).... >>> >>> Best >>> Erick >>> >>> On Mon, Jun 20, 2011 at 8:28 AM, Mark Schoy <hei...@gmx.de> wrote: >>>> Hi, >>>> >>>> if I use a master slave replication in Solr Cloud and the master >>>> crashes, can the slave automatically switch to master mode? >>>> >>>> Or is there another way to index documents after the master is down? >>>> >>>> Thanks. >>>> >>> >> >