New IndexSearcher and autowarming
I would like to have the ability to keep requests from being slowed from new document adds and commits by having a separate index that gets updated. Basically a read-only and an updatable index. After the update index has finished updating with new adds and commits, I'd like to switch the update to the live read-only. At the same time, it would be nice to have the old read-only index become updated with the now live read-only index before I start this update process again. 1. Index1 is live and read-only and doesn't get slowed by updates 2. Index2 is updated with Index1 and gets new adds and commits 3. Index2 gets cache warming 4. Index2 becomes the live index read-only index 5. Index1 gets synced with Index2 so that when these steps start again, the updating is happening on an updated index. I know that this is possible but can't find a simple tutorial on how to do this. By the way, I'm using SolrNet in a windows environment. Thanks, Mike
Re: New IndexSearcher and autowarming
The multicore API (see http://wiki.apache.org/solr/CoreAdmin ) allows you to swap, unload, reload cores. That should allow you to do what you want, -Simon On Fri, Aug 26, 2011 at 11:13 AM, Mike Austin mike.aus...@juggle.comwrote: I would like to have the ability to keep requests from being slowed from new document adds and commits by having a separate index that gets updated. Basically a read-only and an updatable index. After the update index has finished updating with new adds and commits, I'd like to switch the update to the live read-only. At the same time, it would be nice to have the old read-only index become updated with the now live read-only index before I start this update process again. 1. Index1 is live and read-only and doesn't get slowed by updates 2. Index2 is updated with Index1 and gets new adds and commits 3. Index2 gets cache warming 4. Index2 becomes the live index read-only index 5. Index1 gets synced with Index2 so that when these steps start again, the updating is happening on an updated index. I know that this is possible but can't find a simple tutorial on how to do this. By the way, I'm using SolrNet in a windows environment. Thanks, Mike
Re: New IndexSearcher and autowarming
Why doesn't standard replication with auto-warming work for you? You can control how often replication gets triggered by controlling your commit points and/or your replication interval. This seems easier than maintaining cores like your problem statement indicates. Best Erick On Fri, Aug 26, 2011 at 12:56 PM, simon mtnes...@gmail.com wrote: The multicore API (see http://wiki.apache.org/solr/CoreAdmin ) allows you to swap, unload, reload cores. That should allow you to do what you want, -Simon On Fri, Aug 26, 2011 at 11:13 AM, Mike Austin mike.aus...@juggle.comwrote: I would like to have the ability to keep requests from being slowed from new document adds and commits by having a separate index that gets updated. Basically a read-only and an updatable index. After the update index has finished updating with new adds and commits, I'd like to switch the update to the live read-only. At the same time, it would be nice to have the old read-only index become updated with the now live read-only index before I start this update process again. 1. Index1 is live and read-only and doesn't get slowed by updates 2. Index2 is updated with Index1 and gets new adds and commits 3. Index2 gets cache warming 4. Index2 becomes the live index read-only index 5. Index1 gets synced with Index2 so that when these steps start again, the updating is happening on an updated index. I know that this is possible but can't find a simple tutorial on how to do this. By the way, I'm using SolrNet in a windows environment. Thanks, Mike
Re: New IndexSearcher and autowarming
Hi Erick, It might work. I've only worked with solr having one index on one server over a year ago so I might need to just research more about the replication. I am using windows and I remember that replication on windows had some issues with scripts and hard links, however it looks like we have some new good replication features with solr1.4. For now, I wanted to do this on just one windows server since this is my requirement. After your suggestion, I took a little more time to review: http://wiki.apache.org/solr/SolrReplication. So based on what I want to do, would the Replication with MultiCore section be what I need to do? But this wouldn't be a master/slave setup would it since basically I want to swap between two. I guess I could set up 3 indexes on the same server if that's possible to use master/slave in that way, but that might take some more space than I anticipated. Thanks, Mike On Fri, Aug 26, 2011 at 12:08 PM, Erick Erickson erickerick...@gmail.comwrote: Why doesn't standard replication with auto-warming work for you? You can control how often replication gets triggered by controlling your commit points and/or your replication interval. This seems easier than maintaining cores like your problem statement indicates. Best Erick On Fri, Aug 26, 2011 at 12:56 PM, simon mtnes...@gmail.com wrote: The multicore API (see http://wiki.apache.org/solr/CoreAdmin ) allows you to swap, unload, reload cores. That should allow you to do what you want, -Simon On Fri, Aug 26, 2011 at 11:13 AM, Mike Austin mike.aus...@juggle.com wrote: I would like to have the ability to keep requests from being slowed from new document adds and commits by having a separate index that gets updated. Basically a read-only and an updatable index. After the update index has finished updating with new adds and commits, I'd like to switch the update to the live read-only. At the same time, it would be nice to have the old read-only index become updated with the now live read-only index before I start this update process again. 1. Index1 is live and read-only and doesn't get slowed by updates 2. Index2 is updated with Index1 and gets new adds and commits 3. Index2 gets cache warming 4. Index2 becomes the live index read-only index 5. Index1 gets synced with Index2 so that when these steps start again, the updating is happening on an updated index. I know that this is possible but can't find a simple tutorial on how to do this. By the way, I'm using SolrNet in a windows environment. Thanks, Mike