From 160e8980e57ed5310a0e9a65c12db52675fae886 Mon Sep 17 00:00:00 2001
From: Viktor Barzin <viktorbarzin@meta.com>
Date: Sat, 11 Apr 2026 10:30:44 +0100
Subject: [PATCH] perf(immich): restore PostgreSQL vector search optimizations
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

- shared_buffers: 1GB → 2GB (clip_index is 452MB, needs headroom)
- effective_cache_size: 1536MB → 2560MB
- PG memory: 2Gi → 3Gi to support larger shared_buffers
- Add pg_prewarm to shared_preload_libraries with autoprewarm
- First search after restart: 999ms → 25ms
---
 stacks/immich/main.tf | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/stacks/immich/main.tf b/stacks/immich/main.tf
index f272d91c..f9ff83b6 100644
--- a/stacks/immich/main.tf
+++ b/stacks/immich/main.tf
@@ -448,10 +448,10 @@ resource "kubernetes_deployment" "immich-postgres" {
           resources {
             requests = {
               cpu    = "100m"
-              memory = "2Gi"
+              memory = "3Gi"
             }
             limits = {
-              memory = "2Gi"
+              memory = "3Gi"
             }
           }
         }
@@ -461,9 +461,12 @@ resource "kubernetes_deployment" "immich-postgres" {
           command = ["sh", "-c", <<-EOT
             cat > /data/postgresql.override.conf <<'PGCONF'
             # Immich vector search performance tuning
-            shared_buffers = 1024MB
-            effective_cache_size = 1536MB
+            shared_buffers = 2048MB
+            effective_cache_size = 2560MB
             work_mem = 64MB
+            shared_preload_libraries = 'vchord.so, vectors.so, pg_prewarm'
+            pg_prewarm.autoprewarm = on
+            pg_prewarm.autoprewarm_interval = 300
             PGCONF
           EOT
           ]