Re: Deprecating "collection" parameter for routing

2025-01-30 Thread Chris Hostetter
: distributed-search across a number of collections[4]. It's actually a : comma delimited list of collections. AFAICT, this parameter is : undocumented. I want to deprecate this parameter in Solr 9.9, removing it FWIW, It is documented... https://solr.apache.org/guide/solr/latest/deployment-

Re: SIP-7: Merging the new Admin UI (experimental)

2025-01-30 Thread David Smiley
Very exciting! I'm hoping one day to run the UI independently from Solr, as an option. In this way, those of us not yet on Solr 10 could still take advantage of it on our desktops. Maybe this is what you were getting at in your last paragraph. On Thu, Jan 30, 2025 at 2:01 AM Christos Malliaridi

Re: Vector search problems SolR 9.6 (Lucene 9.10) vs. SolR 9.7 (Lucene 9.11)

2025-01-30 Thread Benjamin Trent
Yes, FYI, we found the bug in the kNN query https://github.com/apache/lucene/issues/14180 Basically, threads sharing information back for graph early termination can lead to inconsistency. We should fix this in Lucene. Though I do not know the timeline or the simplicity. Thank you Dr. Andreas Mol

Re: Vector search problems SolR 9.6 (Lucene 9.10) vs. SolR 9.7 (Lucene 9.11)

2025-01-30 Thread Varun Thacker
Benjamin - I think this has to do with Solr 9.7+ using thread executor's for searching. I can take Solr 9.7 or Solr 9.8 and just undo this one line in SolrIndexSearcher

Re: [JENKINS] Solr-main-Linux (64bit/hotspot/jdk-11.0.21) - Build # 19896 - Still Unstable!

2025-01-30 Thread David Smiley
BTW, I must compliment you on your great investigative skills to identify that as the root cause! Truly impressive! On Wed, Jan 29, 2025 at 3:01 PM Houston Putman wrote: > So after looking into this a bit, it looks like the versions are being > returned in a different order. > > This reproduces

Re: [JENKINS] Solr-main-Linux (64bit/hotspot/jdk-11.0.21) - Build # 19896 - Still Unstable!

2025-01-30 Thread David Smiley
Disclaimer: update ordering is not something I know much about. But... I wish there was a total order of all updates, so that all updates to the log were ordered by design. If that were so, this wouldn't be a bug here and other complexities would likely go away. I'm thinking of this: https://gi