John Leach

Binary Points have been optimized for deletes in the new versions of Lucene. From the javadoc… /** * Create a query for matching an exact {@link BytesRef} value. * NOTE:...

Path... ``` Stack: [0x000070000d0d9000,0x000070000d1d9000], sp=0x000070000d1d21f0, free space=996k Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code) V [libjvm.dylib+0x31da5b] jni_SetStaticIntField+0xa0 C [libjansi.jnilib+0xf98] Java_org_fusesource_jansi_internal_CLibrary_init+0x108 j org.fusesource.jansi.internal.CLibrary.init()V+0 j org.fusesource.jansi.internal.CLibrary.()V+21 v ~StubRoutines::call_stub V...

Cleanup Fake KubeClient Construction #1323

1

Cleaning up repetitive calls in construction of fake KubeClient.

### What would you like to be added/changed? Remove some repetitive calls and setup (nit).

Instead of doing block based storedFields, we should support key based storedFields to optimize the read amplification from FoundationDB.

Goal would be for readblock to produce not just the file it is reading but the reference inside the file it is attempting to read after the slice.

Reference Mail Implementation for the Lucene Package FoundationDB#2152

6

This will be used as a tool to reason about the Lucene implementations scalability with regards to optimizations in Lucene.

LuceneOptimizedCodec is currently created per indexWriter.

Add cache for Lucene schema information

1

Add a cache similar to block cache... The reads seem to come up on the metrics even if using the RYW cache.

The metric was created for read schema but never actually made it into the call. :(