The test simply ensures that single node queries match multi node queries. I dont think you want qt.shards=/select necessarily. Not unless it has the component in it along with whatever handler you are using to do the top level distrib request.
At some point I'll set it up and add some doc to the wiki. - mark On Dec 13, 2012, at 5:36 PM, Shawn Heisey <[email protected]> wrote: > On 12/13/2012 1:35 PM, Mark Miller wrote: >> Take a look at the test - that may help. For spellcheck and this thing I >> think you pretty much have to use shards.qt or put the component in your >> standard handler as well. Have you done anything along those lines? >> >> The test shows a sample config and queries if you take a look at the diff. >> >> Let me know. > > I tried adding shards.qt=/select and there was no change.I'm not including > the qt paramter, but since I don't have the dispatcher turned on, I am not > expecting that to be a problem. > > Looking at the test, I can't figure out how it passes - if I understand it > correctly, you tell it that you're looking for document number 17 to be in > the MLT results, then you do three very different queries, one of which > intentionally doesn't match any of your documents. I will admit that my > review of the base class that the test extends was not extensive, so I am > probably interpreting it badly. > > Just to be sure I hadn't screwed something up when I updated my working copy, > I did a brand new checkout of branch_4x and applied the patch for SOLR-4143. > Thanks to you and Hoss committing some of my other work, that's the only > patch that I am actively using. Putting that .war into place, it still > didn't work. > > I have not reindexed since upgrading Solr, but I haven't seen any errors in > my logs relating to the index format changing. I will wipe my index > directory, restart, and do a reindex. Next test in about three hours! > > Thanks, > Shawn > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
