RE: disabling QueryElevationComponent
Walter Underwood [mailto:wun...@wunderwood.org] writes: >On Mar 5, 2012, at 1:16 PM, Welty, Richard wrote: >> Walter Underwood [mailto:wun...@wunderwood.org] writes: >>> You may be able to have unique keys. At Netflix, I found that there were >>> collisions between the movie IDs and the person IDs. So, I put an 'm' at >>> the beginning of each movie ID and a 'p' at the beginning of each person >>> ID. Like magic, I had unique IDs. >> did you do this with a transformer at index time, or in some other manner? >SQL should be able to do this, though it might not be portable. For MySQL is > it something like: >select > concat('m', movie_id) as id, ok, thanks. i know how to do that in postgresql... richard
Re: disabling QueryElevationComponent
On Mar 5, 2012, at 1:16 PM, Welty, Richard wrote: > Walter Underwood [mailto:wun...@wunderwood.org] writes: > >> You may be able to have unique keys. At Netflix, I found that there were >> collisions between >the movie IDs and the person IDs. So, I put an 'm' at >> the beginning of each movie ID and a >'p' at the beginning of each person >> ID. Like magic, I had unique IDs. > > did you do this with a transformer at index time, or in some other manner? I had custom code for indexing; this was in Solr 1.3. SQL should be able to do this, though it might not be portable. For MySQL is it something like: select concat('m', movie_id) as id, ... wunder -- Walter Underwood wun...@wunderwood.org
RE: disabling QueryElevationComponent
Walter Underwood [mailto:wun...@wunderwood.org] writes: >You may be able to have unique keys. At Netflix, I found that there were >collisions between >the movie IDs and the person IDs. So, I put an 'm' at the >beginning of each movie ID and a >'p' at the beginning of each person ID. Like >magic, I had unique IDs. did you do this with a transformer at index time, or in some other manner? >You should be able to disable the query elevation stuff by removing it from >your >solrconfig.xml. the documentation certainly implies this, which is why i'm baffled. i see no reason why removing the config should trigger the multiple core error when i only have the default setup with one core. richard
Re: disabling QueryElevationComponent
You may be able to have unique keys. At Netflix, I found that there were collisions between the movie IDs and the person IDs. So, I put an 'm' at the beginning of each movie ID and a 'p' at the beginning of each person ID. Like magic, I had unique IDs. You should be able to disable the query elevation stuff by removing it from your solrconfig.xml. wunder Walter Underwood wun...@wunderwood.org On Mar 5, 2012, at 1:09 PM, Welty, Richard wrote: > i googled and found numerous references to this, but no answers that went to > my specific issues. > > i have a solr 3.5.0 server set up that needs to index several different > document types, there is no common unique key field. so i can't use the > uniqueKey declaration and need to disable the QueryElevationComponent. > > when i set this up with the uniqueKey, i mapped ids from the various database > tables to the id key temporarily just to get things working, but the results > from queries are showing me i really have to get rid of that hack. so i > commented out uniqueKey in schema.xml and commented out the > QueryElevationComponent searchComponent and the associated requestHandler in > solrconfig.xml > > when i restart solr and go to the solr/admin/dataimport.jsp page to test, i > get the 'missing core name in path' error. > > so what further configuration changes are required to disable > QueryElevationComponent? > > thanks, > richard --