I don’t feel comfortable with waiting another week and then including a rushed patch release.
I think it’s much better for everyone involved to just accept the state of things as they are - BUT document it well AND provide a clear procedure on what bits to replace once an official release becomes available. Cheers, Andrzej > On 14 Oct 2026, at 02:22, Rahul Goswami <[email protected]> wrote: > > Update: After initial reconsideration, the maintainers are willing to > provide a quicker turnaround with a patch on top of cuvs-lucene 26.10 > (latest released) compatible with Lucene 10.4. It won’t be a formal release > but they can make it available by mid next week. > Is this an option we are willing to consider for 10.1 ? > > The alternative of course is to have a note in the upgrade docs for now and > patch this when the official 26.12 becomes available later this year. > > - Rahul > > On Tue, Oct 13, 2026 at 1:38 PM Rahul Goswami <[email protected]> wrote: > >>> "They would be able to upgrade to Solr-10.1 just >>> as soon as a Solr-10.1 compatable version of cuvs was available in that >>> new/different repo" >> How would they know? Are there releases for sandbox modules? Also, if I >> understand correctly that would require users to build the jars themselves, >> correct? >> >> Also as expressed on the RC2 thread, we don't have to delay the release - >> I'll add a note (to >> solr/solr-ref-guide/modules/upgrade-notes/pages/major-changes-in-solr-10.adoc >> ?) >> about the breakage, and availability of a subsequent fix in the future. I >> am told we are going to have a 10.1 RC3 anyway to accommodate other fixes >> and possibly the Jetty upgrade. >> >> Reason I am not entirely sold on the sandbox idea is we are talking about >> abruptly telling production users to monitor the sandbox based on a one >> time occurence despite a known timeline for a fix...at a time when other >> popular Lucene based search engines today ship with GPU acceleration >> capabilities for vector search. Probably if this becomes a repeating >> concern we revisit this option then? >> >> @Eric Pugh <[email protected]> That's a good question. I >> did enquire about the possibility of expediting the fix for Lucene 10.4, >> however it seems it entails more work than could be accommodated in a short >> window of a week or two. But they are looking to release cuvs-lucene 26.12 >> in December which will have support through Lucene 10.5 (atleast). The >> planned cadence I have been made aware of is a cuvs-lucene release every >> two months starting December compatible with the latest Lucene version >> released as of the start of the release cycle of cuvs-lucene. >> >> Thanks, >> Rahul >> >> On Tue, Oct 13, 2026 at 11:53 AM Eric Pugh <[email protected]> wrote: >> >>> >>> >>> On 2026/10/13 15:31:57 Anshum Gupta wrote: >>>> +1 to Hoss’s idea of moving this to the sandbox. >>>> >>>> >>>> Anshum Gupta >>>> >>>> On Tue, Oct 13, 2026 at 00:13 Jan Høydahl <[email protected]> wrote: >>>> >>>>> +1 to move to sandbox. Tiny feature should not hold us hostage >>>>> >>>>> Jan >>>>> >>>>>> 12. okt. 2026 kl. 23:12 skrev Chris Hostetter < >>> [email protected] >>>>>> : >>>>>> >>>>>> >>>>>> The bugs found with solr/modules/cuvs in the Solr 10.1.0 RCs -- due >>> to >>>>> the >>>>>> module depending on a thrid-party dependency that is tightly couple >>> with >>>>>> the Lucene version, but released by a completely different >>> organization >>>>> -- >>>>>> is really just a new variation on the fundemental the concerns I >>> raise >>>>>> when cuvs was first added to Solr... >>>>>> >>>>>> https://issues.apache.org/jira/browse/SOLR-17892 >>>>>> >>>>>> I get that cuvs is a valuable feature to have, for some people, on >>> some >>>>>> hardware, but I really don't think it makes sense for it to be part >>> of >>>>> the >>>>>> official Solr release -- not when it's depenencies are so tightly >>>>>> intertwined (but loosly coupled) to the specific Lucene version at a >>>>> very >>>>>> low level. >>>>>> >>>>>> As Rahul indicates -- we are now facing the exact choice I warmed >>> about >>>>> a >>>>>> year ago... >>>>>> >>>>>>>> In the future we (Solr) may get stuck between a rock and a hard >>> place >>>>>>>> when attempting a Lucene upgrades ... we will be forced to >>> choose: >>>>>>>> >>>>>>>> - Let the Lucene upgrade be blocked indefinitely until a >>>>>>>> third-party project ( cuvs-lucene ) gets around to a >>>>>>>> compatible release. >>>>>>>> - Stop supporting modules/cuvs in what may be a MINOR Solr >>>>>>>> release, potentially screwing our users who have built indexes >>>>>>>> using that module >>>>>> >>>>>> >>>>>> I really think we need to move cuvs to either the solr-sandbox, or >>> some >>>>>> new purpose specific 'solr-cuvs' git repo, where it can be released >>> on >>>>>> it's own timeframe, dependent on when/how/if the upstream >>> third-party >>>>>> project (cuvs-lucene) has compatible releases for the Lucene >>> version we >>>>>> are using. >>>>>> >>>>>> >>>>>> People who use Solr and do *not* use modules/cuvs should not have to >>>>>> suffer delays in getting iportant bug fixes in Solr code -- or >>> getting a >>>>>> Solr release with a Lucene version that's already outdated -- just >>>>> because >>>>>> we don't want to release a Solr version with a "broken" >>> modules/cuvs. >>>>>> >>>>>> >>>>>> -Hoss >>>>>> http://www.lucidworks.com/ >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> On Sun, 11 Oct 2026, Rahul Goswami wrote: >>>>>> >>>>>> : Date: Sun, 11 Oct 2026 17:08:02 -0400 >>>>>> : From: Rahul Goswami <[email protected]> >>>>>> : Reply-To: [email protected] >>>>>> : To: [email protected] >>>>>> : Subject: Re: [VOTE] Release Solr 10.1.0 RC2 >>>>>> : >>>>>> : Tried the RC. The cuvs module seems to be broken. Indexing fails >>> on a >>>>> GPU >>>>>> : enabled machine due to API incompatibility between cuvs-lucene and >>>>> Lucene >>>>>> : 10.4. >>>>>> : Currently no available compatible cuvs-lucene version exists for >>> Lucene >>>>>> : 10.4 and is expected to be available by December this year. >>>>>> : >>>>>> : If there is agreement, I could spend the effort on rolling Solr >>> back to >>>>>> : Lucene 10.3.2 where I have validated cuvs to work (which was also >>> the >>>>>> : Lucene version we were on prior to 10.4). >>>>>> : >>>>>> : Thoughts? >>>>>> : >>>>>> : 2026-10-10 20:57:58.804 WARN (qtp166454155-71) [] >>>>> o.e.j.e.s.ServletChannel >>>>>> : /solr/vectors/update >>>>>> : *java.lang.NoSuchMethodError: 'int[] >>>>>> : >>>>> >>> org.apache.lucene.util.hnsw.HnswGraph$NodesIterator.getSortedNodes(org.apache.lucene.util.hnsw.HnswGraph$NodesIterator)'* >>>>>> : at >>>>>> : >>>>> >>> com.nvidia.cuvs.lucene.Lucene99AcceleratedHNSWVectorsWriter.writeGraph(Lucene99AcceleratedHNSWVectorsWriter.java:444) >>>>>> : at >>>>>> : >>>>> >>> com.nvidia.cuvs.lucene.Lucene99AcceleratedHNSWVectorsWriter.writeFieldInternal(Lucene99AcceleratedHNSWVectorsWriter.java:230) >>>>>> : at >>>>>> : >>>>> >>> com.nvidia.cuvs.lucene.Lucene99AcceleratedHNSWVectorsWriter.writeField(Lucene99AcceleratedHNSWVectorsWriter.java:502) >>>>>> : at >>>>>> : >>>>> >>> com.nvidia.cuvs.lucene.Lucene99AcceleratedHNSWVectorsWriter.flush(Lucene99AcceleratedHNSWVectorsWriter.java:494) >>>>>> : at >>>>>> : >>>>> >>> org.apache.lucene.codecs.perfield.PerFieldKnnVectorsFormat$FieldsWriter.flush(PerFieldKnnVectorsFormat.java:120) >>>>>> : at >>>>>> : >>>>> >>> org.apache.lucene.index.VectorValuesConsumer.flush(VectorValuesConsumer.java:76) >>>>>> : at >>> org.apache.lucene.index.IndexingChain.flush(IndexingChain.java:305) >>>>>> : at >>>>>> : >>>>> >>> org.apache.lucene.index.DocumentsWriterPerThread.flush(DocumentsWriterPerThread.java:456) >>>>>> : at >>>>> >>> org.apache.lucene.index.DocumentsWriter.doFlush(DocumentsWriter.java:502) >>>>>> : at >>>>>> : >>>>> >>> org.apache.lucene.index.DocumentsWriter.maybeFlush(DocumentsWriter.java:456) >>>>>> : at >>>>>> : >>>>> >>> org.apache.lucene.index.DocumentsWriter.flushAllThreads(DocumentsWriter.java:649) >>>>>> : at >>>>>> : >>>>> >>> org.apache.lucene.index.IndexWriter.prepareCommitInternal(IndexWriter.java:3712) >>>>>> : at >>>>> >>> org.apache.lucene.index.IndexWriter.commitInternal(IndexWriter.java:4160) >>>>>> : at >>> org.apache.lucene.index.IndexWriter.commit(IndexWriter.java:4122) >>>>>> : at >>>>>> : >>>>> >>> org.apache.solr.update.DirectUpdateHandler2.commit(DirectUpdateHandler2.java:872) >>>>>> : at >>>>>> : >>>>> >>> org.apache.solr.update.processor.RunUpdateProcessorFactory$RunUpdateProcessor.processCommit(RunUpdateProcessorFactory.java:100) >>>>>> : at >>>>>> : >>>>> >>> org.apache.solr.update.processor.UpdateRequestProcessor.processCommit(UpdateRequestProcessor.java:66) >>>>>> : at >>>>>> : >>>>> >>> org.apache.solr.update.processor.DistributedUpdateProcessor.doLocalCommit(DistributedUpdateProcessor >>>>>> : >>>>>> : - Rahul >>>>>> : >>>>>> : On Sat, Oct 10, 2026 at 5:57 PM Jan Høydahl <[email protected]> >>> wrote: >>>>>> : >>>>>> : > The changelog lists "Dependency Upgrades (61 changes)" >>>>>> : > I thought we were to hand edit and de-duplicate that list. >>>>>> : > >>>>>> : > PS: Jetty 12.1.14 hits branch_10x soon, hope it can make it if >>>>> there's a >>>>>> : > respin. >>>>>> : > >>>>>> : > Jan >>>>>> : > >>>>>> : > > 10. okt. 2026 kl. 15:48 skrev Andrzej Białecki <[email protected] >>>> : >>>>>> : > > >>>>>> : > > Please vote for release candidate 2 for Solr 10.1.0 >>>>>> : > > >>>>>> : > > The artifacts can be downloaded from: >>>>>> : > > >>>>>> : > >>>>> >>> https://dist.apache.org/repos/dist/dev/solr/solr-10.1.0-RC2-rev-dfda5e1d4821958e60d13aa5ce3a0d4992319ed1 >>>>>> : > > >>>>>> : > > You can run the smoke tester directly with this command: >>>>>> : > > >>>>>> : > > python3 -u dev-tools/scripts/smokeTestRelease.py \ >>>>>> : > > >>>>>> : > >>>>> >>> https://dist.apache.org/repos/dist/dev/solr/solr-10.1.0-RC2-rev-dfda5e1d4821958e60d13aa5ce3a0d4992319ed1 >>>>>> : > > >>>>>> : > > You can build a release-candidate of the official docker >>> images >>>>> (full & >>>>>> : > slim) using the following command: >>>>>> : > > >>>>>> : > > SOLR_DOWNLOAD_SERVER= >>>>>> : > >>>>> >>> https://dist.apache.org/repos/dist/dev/solr/solr-10.1.0-RC2-rev-dfda5e1d4821958e60d13aa5ce3a0d4992319ed1/solr >>>>>> : > && \ >>>>>> : > > docker build >>>>>> : > $SOLR_DOWNLOAD_SERVER/10.1.0/docker/Dockerfile.official-full \ >>>>>> : > > --build-arg SOLR_DOWNLOAD_SERVER=$SOLR_DOWNLOAD_SERVER \ >>>>>> : > > -t solr-rc:10.1.0-2 && \ >>>>>> : > > docker build >>>>>> : > $SOLR_DOWNLOAD_SERVER/10.1.0/docker/Dockerfile.official-slim \ >>>>>> : > > --build-arg SOLR_DOWNLOAD_SERVER=$SOLR_DOWNLOAD_SERVER \ >>>>>> : > > -t solr-rc:10.1.0-2-slim >>>>>> : > > >>>>>> : > > The vote will be open for 72 hours (not including the Sunday >>> and >>>>>> : > Columbus Day), i.e. until 2026-10-15 15:00 UTC. >>>>>> : > > >>>>>> : > > [ ] +1 approve >>>>>> : > > [ ] +0 no opinion >>>>>> : > > [ ] -1 disapprove (and reason why) >>>>>> : > > >>>>>> : > > Here is my +1 >>>>>> : > > ---- >>>>>> : > > >>>>>> : > > Cheers, >>>>>> : > > Andrzej >>>>>> : > > >>>>> --------------------------------------------------------------------- >>>>>> : > > 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] >>>>>> : > >>>>>> : > >>>>>> : >>>>>> >>>>>> >>> --------------------------------------------------------------------- >>>>>> 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] >>>>> >>>>> >>>> >>> What about highlighting to that community that we need a 10.4 release >>> ASAP, and seeing if that changes their timeline. Judinging from the github >>> issue, they are filling in a number of missing verion releases, which of >>> couse makes it more work then just cutting a 10.4 release. If they >>> committed to releasing a 10.4 in the next week, would that change things >>> for us? >>> >>> --------------------------------------------------------------------- >>> 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]
