chore: log and tweak invalidation

Author: danielroe

server/api/clusters/[...repo].ts

@@ -36,8 +36,10 @@ export default defineCachedEventHandler(async (event) => {
   const backlog = [...issues]
   const embeddings: number[][] = []
 
+  let b = 1
   do {
     const batch = backlog.splice(0, 100)
+    console.log('fetching batch', b++)
     embeddings.push(...await Promise.all(batch.map(async issue => getEmbeddingsForIssue(event, issue))))
   } while (backlog.length)
 
@@ -56,4 +58,5 @@ export default defineCachedEventHandler(async (event) => {
   maxAge: 60 * 60 * 1000,
   staleMaxAge: 60 * 60 * 1000,
   shouldBypassCache: event => getHeader(event, 'force') === 'true',
+  shouldInvalidateCache: event => getHeader(event, 'force') === 'true',
 })

server/api/issues/[...repo].ts

@@ -48,4 +48,5 @@ export default defineCachedEventHandler(async (event) => {
   maxAge: 60 * 60 * 1000,
   staleMaxAge: 60 * 60 * 1000,
   shouldBypassCache: event => getHeader(event, 'force') === 'true',
+  shouldInvalidateCache: event => getHeader(event, 'force') === 'true',
 })

server/utils/cluster.ts

@@ -37,7 +37,7 @@ export function clusterEmbeddings(issues: Issue[], embeddings: number[][]) {
       return { ...issue, avgSimilarity: Math.floor(100 * totalSimilarity / (chunk.length - 1)) / 100 }
     })
 
-    const similarityThreshold = 0.75
+    const similarityThreshold = 0.30 // 0.75
 
     sortedChunk = sortedChunk.filter(i => i.avgSimilarity >= similarityThreshold).sort((a, b) => b.avgSimilarity - a.avgSimilarity)