Re: manual background re-indexing
I understand that multicores allows met o create two cores which are just normal solr-homes. That's easy. However I'm only interested to the reload command and, in particular, to reload the underlying index segment files. Do I understand correctly that the following is what I should do? - still make my setting multicore and get the core-admin requesthandler to work, even with one core - attempt the reload with a change of solrconfig or schema - do the reload of data by changing the index-segment-path in the config as an example of the above thanks to clarify paul Le 29 avr. 2011 à 03:12, Erick Erickson a écrit : You simply create two cores. One in solr/cores/core1 and another in solr/cores/core2 They each have a separate conf and data directory,and the index in in core#/data/index. Really, its' just introducing one more level. You can experiment just by configuring a core and copying your index to solr/cores/yourcore/data/index. After, of course, configuring Solr.xml to understand cores.
Re: manual background re-indexing
I'm a little confused about your use of the word reload, see below. But if this doesn't clarify things, or your experiments give strange results, could you please start posting sample configurations so we can be sure we're talking about the same thing? Best Erick On Fri, Apr 29, 2011 at 5:15 AM, Paul Libbrecht p...@hoplahup.net wrote: I understand that multicores allows met o create two cores which are just normal solr-homes. That's easy. However I'm only interested to the reload command and, in particular, to reload the underlying index segment files. Do I understand correctly that the following is what I should do? - still make my setting multicore and get the core-admin requesthandler to work, even with one core Great place to start. - attempt the reload with a change of solrconfig or schema Should work find. A reload will cause that core to understand the changes to the schema - do the reload of data by changing the index-segment-path in the config as an example of the above Here's where I'm confused. Changing the index-segment-path in config to point where? I guess you *could* point at your old Solr index directory for, say, core1. But the normal path would be solr home/cores/core1/. Under that directory you should have a bin, conf, and data just like you had under solr home. Reloading doesn't have much to do with the actual contents of the .../data directory, if you relocate stuff you'd have to copy those three directories to the place that corresponds to the path in solrconfig.xml I've never had one core point to the solr home directory and other cores point to the normal place, so I can't speak from experience but it seems to me that this should work. thanks to clarify paul Le 29 avr. 2011 à 03:12, Erick Erickson a écrit : You simply create two cores. One in solr/cores/core1 and another in solr/cores/core2 They each have a separate conf and data directory,and the index in in core#/data/index. Really, its' just introducing one more level. You can experiment just by configuring a core and copying your index to solr/cores/yourcore/data/index. After, of course, configuring Solr.xml to understand cores.
manual background re-indexing
Hello list, I am planning to implement a setup, to be run on unix scripts, that should perform a full pull-and-reindex in a background server and index then deploy that index. All should happen on the same machine. I thought the replication methods would help me but they seem to rather solve the issues of distribution while, what I need, is only the ability to: - suspend the queries - swap the directories with the new index - close all searchers - reload and warm-up the searcher on the new index Is there a part of the replication utilities (http or unix) that I could use to perform the above tasks? I intend to do this on occasion... maybe once a month or even less. Is reload the right term to be used? paul
Re: manual background re-indexing
Hi Paul Would a multi-core set up and the swap command do what you want it to do? http://wiki.apache.org/solr/CoreAdmin Shaun On 28 April 2011 12:49, Paul Libbrecht p...@hoplahup.net wrote: Hello list, I am planning to implement a setup, to be run on unix scripts, that should perform a full pull-and-reindex in a background server and index then deploy that index. All should happen on the same machine. I thought the replication methods would help me but they seem to rather solve the issues of distribution while, what I need, is only the ability to: - suspend the queries - swap the directories with the new index - close all searchers - reload and warm-up the searcher on the new index Is there a part of the replication utilities (http or unix) that I could use to perform the above tasks? I intend to do this on occasion... maybe once a month or even less. Is reload the right term to be used? paul
Re: manual background re-indexing
Just where to do I put the new index data with such a command? Simply replacing the segment files appears dangerous to me. Also, what is the best practice to move from single-core to multi-core? My current set-up is single-core, do I simply need to add a solr.xml in my solr-home and one core1 directory with the data that was there previously? paul Le 28 avr. 2011 à 14:04, Shaun Campbell a écrit : Hi Paul Would a multi-core set up and the swap command do what you want it to do? http://wiki.apache.org/solr/CoreAdmin Shaun On 28 April 2011 12:49, Paul Libbrecht p...@hoplahup.net wrote: Hello list, I am planning to implement a setup, to be run on unix scripts, that should perform a full pull-and-reindex in a background server and index then deploy that index. All should happen on the same machine. I thought the replication methods would help me but they seem to rather solve the issues of distribution while, what I need, is only the ability to: - suspend the queries - swap the directories with the new index - close all searchers - reload and warm-up the searcher on the new index Is there a part of the replication utilities (http or unix) that I could use to perform the above tasks? I intend to do this on occasion... maybe once a month or even less. Is reload the right term to be used? paul
Re: manual background re-indexing
It would probable be safest just to set up a separate system as multi-core from the start, get the process working and then either use the new machine or copy the whole setup to the production machine. Best Erick On Thu, Apr 28, 2011 at 8:49 AM, Paul Libbrecht p...@hoplahup.net wrote: Just where to do I put the new index data with such a command? Simply replacing the segment files appears dangerous to me. Also, what is the best practice to move from single-core to multi-core? My current set-up is single-core, do I simply need to add a solr.xml in my solr-home and one core1 directory with the data that was there previously? paul Le 28 avr. 2011 à 14:04, Shaun Campbell a écrit : Hi Paul Would a multi-core set up and the swap command do what you want it to do? http://wiki.apache.org/solr/CoreAdmin Shaun On 28 April 2011 12:49, Paul Libbrecht p...@hoplahup.net wrote: Hello list, I am planning to implement a setup, to be run on unix scripts, that should perform a full pull-and-reindex in a background server and index then deploy that index. All should happen on the same machine. I thought the replication methods would help me but they seem to rather solve the issues of distribution while, what I need, is only the ability to: - suspend the queries - swap the directories with the new index - close all searchers - reload and warm-up the searcher on the new index Is there a part of the replication utilities (http or unix) that I could use to perform the above tasks? I intend to do this on occasion... maybe once a month or even less. Is reload the right term to be used? paul
Rép : Re: manual background re-indexing
It would probable be safest just to set up a separate system as multi-core from the start, get the process working and then either use the new machine or copy the whole setup to the production machine. On Thu, Apr 28, 2011 at 8:49 AM, Paul Libbrecht p...@hoplahup.net wrote: Just where to do I put the new index data with such a command? Simply replacing the segment files appears dangerous to me.Any idea where I should put the data directory before calling the reload command?paul
Re: Rép : Re: manual background re-indexing
You simply create two cores. One in solr/cores/core1 and another in solr/cores/core2 They each have a separate conf and data directory,and the index in in core#/data/index. Really, its' just introducing one more level. You can experiment just by configuring a core and copying your index to solr/cores/yourcore/data/index. After, of course, configuring Solr.xml to understand cores. Best Erick On Thu, Apr 28, 2011 at 7:27 PM, Paul Libbrecht p...@hoplahup.net wrote: It would probable be safest just to set up a separate system as multi-core from the start, get the process working and then either use the new machine or copy the whole setup to the production machine. On Thu, Apr 28, 2011 at 8:49 AM, Paul Libbrecht p...@hoplahup.net wrote: Just where to do I put the new index data with such a command? Simply replacing the segment files appears dangerous to me. Any idea where I should put the data directory before calling the reload command? paul