perf: run and external_users queries (#459)

lunary-ai · Aug 2, 2024 · 18e0605 · 18e0605
1 parent 1dd2781
commit 18e0605
Show file tree

Hide file tree

Showing 4 changed files with 96 additions and 46 deletions.
diff --git a/packages/backend/src/api/v1/runs/index.ts b/packages/backend/src/api/v1/runs/index.ts
@@ -191,15 +191,15 @@ function formatRun(run: any) {
         })
       }
     }
+    // TODO: put in an array nammed enrichment instead
+    for (let evaluationResult of run.evaluationResults || []) {
+      formattedRun[`enrichment-${evaluationResult.evaluatorId}`] =
+        evaluationResult
+    }
   } catch (error) {
     console.error(error)
   }
 
-  // TODO: put in an array nammed enrichment instead
-  for (let evaluationResult of run.evaluationResults || []) {
-    formattedRun[`enrichment-${evaluationResult.evaluatorId}`] =
-      evaluationResult
-  }
   return formattedRun
 }
 
@@ -243,48 +243,61 @@ runs.get("/", async (ctx: Context) => {
   }
 
   const rows = await sql`
+    with runs as (
+        select
+            r.*,
+            eu.id as user_id,
+            eu.external_id as user_external_id,
+            eu.created_at as user_created_at,
+            eu.last_seen as user_last_seen,
+            eu.props as user_props,
+            t.slug as template_slug,
+            rpfc.feedback as parent_feedback
+        from
+            public.run r
+            left join external_user eu on r.external_user_id = eu.id
+            left join run_parent_feedback_cache rpfc on r.id = rpfc.id
+            left join template_version tv on r.template_version_id = tv.id
+            left join template t on tv.template_id = t.id
+            left join evaluation_result_v2 er on r.id = er.run_id
+            left join evaluator e on er.evaluator_id = e.id
+        where
+          r.project_id = ${projectId}
+          ${parentRunCheck}
+          and (${filtersQuery})
+        order by
+            ${sql.unsafe(orderByClause)}  
+        limit ${exportType ? 10000 : Number(limit)}
+        offset ${Number(page) * Number(limit)}
+    ),
+    evaluation_results as (
+        select
+            r.id,
+            coalesce(array_agg(
+              jsonb_build_object(
+                  'evaluatorName', e.name,
+                  'evaluatorSlug', e.slug,
+                  'evaluatorType', e.type,
+                  'evaluatorId', e.id,
+                  'result', er.result, 
+                  'createdAt', er.created_at,
+                  'updatedAt', er.updated_at
+              )
+          ) filter (where er.run_id is not null), '{}') as evaluation_results
+        from
+            runs r
+        left join evaluation_result_v2 er on r.id = er.run_id
+        left join evaluator e on er.evaluator_id = e.id
+        group by r.id
+    )
     select
-      r.*,
-      eu.id as user_id,
-      eu.external_id as user_external_id,
-      eu.created_at as user_created_at,
-      eu.last_seen as user_last_seen,
-      eu.props as user_props,
-      t.slug as template_slug,
-      rpfc.feedback as parent_feedback,
-      coalesce(array_agg(
-          jsonb_build_object(
-              'evaluatorName', e.name,
-              'evaluatorSlug', e.slug,
-              'evaluatorType', e.type,
-              'evaluatorId', e.id,
-              'result', er.result, 
-              'createdAt', er.created_at,
-              'updatedAt', er.updated_at
-          )
-      ) filter (where er.run_id is not null), '{}') as evaluation_results
+        r.*,
+        er.evaluation_results
     from
-      public.run r
-      left join external_user eu on r.external_user_id = eu.id
-      left join run_parent_feedback_cache rpfc on r.id = rpfc.id
-      left join template_version tv on r.template_version_id = tv.id
-      left join template t on tv.template_id = t.id
-      left join evaluation_result_v2 er on r.id = er.run_id 
-      left join evaluator e on er.evaluator_id = e.id
-    where
-      r.project_id = ${projectId}
-      ${parentRunCheck}
-      and (${filtersQuery})
-    group by
-      r.id,
-      eu.id,
-      t.slug,
-      rpfc.feedback
-    order by
-       ${sql.unsafe(orderByClause)}  
-    limit ${exportType ? 10000 : Number(limit)}
-    offset ${Number(page) * Number(limit)}
+      runs r
+      left join evaluation_results er on r.id = er.id;
   `
+
   const runs = rows.map(formatRun)
 
   if (exportType) {

diff --git a/packages/backend/src/utils/db.ts b/packages/backend/src/utils/db.ts
@@ -13,8 +13,8 @@ const sql = postgres(process.env.DATABASE_URL!, {
   connection: {
     application_name: `backend-${isProduction ? "production" : "development"}-${new Date().getTime()}`,
   },
-  // debug: process.env.DEBUG ? debugFn : () => {},
-  // onnotice: process.env.DEUG ? console.log : () => {},
+  debug: process.env.LUNARY_DEBUG ? debugFn : () => {},
+  onnotice: process.env.LUNARY_DEBUG ? console.log : () => {},
 })
 
 function debugFn(

diff --git a/packages/db/0030.sql b/packages/db/0030.sql
@@ -0,0 +1,27 @@
+drop index if exists run_created_at_idx1;
+drop index if exists run_created_at_idx3;
+drop index if exists run_cost_idx;
+drop index if exists run_cost_idx1;
+drop index if exists run_created_at_project_id_idx;
+drop index if exists run_expr_idx1;
+drop index if exists run_expr_idx2;
+drop index if exists run_type_external_user_id_idx;
+drop index if exists run_type_idx;
+drop index if exists run_type_parent_run_id_idx;
+
+create index on run (project_id, created_at);
+create index on run (project_id, cost);
+create index on run (project_id, external_user_id);
+create index on run (project_id, (error is not null));
+
+create index on run (project_id, created_at desc nulls last);
+create index on run (project_id, created_at asc nulls last);
+create index on run (project_id, duration desc nulls last);
+create index on run (project_id, duration asc nulls last);
+create index on run (project_id, (prompt_tokens + completion_tokens) desc nulls last);
+create index on run (project_id, (prompt_tokens + completion_tokens) asc nulls last);
+create index on run (project_id, cost desc nulls last);
+create index on run (project_id, cost asc nulls last);
+
+
+analyze;
diff --git a/packages/db/0031.sql b/packages/db/0031.sql
@@ -0,0 +1,10 @@
+drop index if exists external_user_project_id_last_seen_idx;
+drop index if exists external_user_last_seen_idx;
+drop index if exists external_user_created_at_idx;
+
+create index on external_user (project_id, created_at desc nulls last);
+create index on external_user (project_id, created_at asc nulls last);
+create index on external_user (project_id, last_seen desc nulls last);
+create index on external_user (project_id, last_seen asc nulls last);
+
+analyze;