[VectorSearch] Update opensearch and update cluster configuration for…

… memory management
Elastic-Suite · May 24, 2024 · 95b824a · 95b824a
1 parent ec82325
commit 95b824a
Show file tree

Hide file tree

Showing 2 changed files with 23 additions and 17 deletions.
diff --git a/api/Dockerfile b/api/Dockerfile
@@ -6,7 +6,9 @@
 # https://docs.docker.com/engine/reference/builder/#understand-how-arg-and-from-interact
 ARG PHP_VERSION=8.1
 ARG CADDY_VERSION=2.6.4
-ARG OPENSEARCH2_VERSION=2.13.0
+# On opensearch update, check a fix for this issue has been released : https://github.com/opensearch-project/ml-commons/issues/2308
+# and if so, check if we can re-enable test on multiple model and remove the configuration of the memory_threshold of the cluster.
+ARG OPENSEARCH2_VERSION=2.14.0
 ARG ELASTIC7_VERSION=7.16.3
 ARG ELASTIC8_VERSION=8.10.4
 # ARG XDEBUG_VERSION=3.1.2

diff --git a/docker-compose.yml b/docker-compose.yml
@@ -131,15 +131,17 @@ services:
       target: gally_opensearch2
     # restart: unless-stopped
     environment:
-      - cluster.name=os-docker-cluster     # Search cluster name
-      - node.name=opensearch-node-data     # Name the node that will run in this container
-      - bootstrap.memory_lock=true         # Disable JVM heap memory swapping
-      - "OPENSEARCH_JAVA_OPTS=-Xms512m -Xmx512m" # Set min and max JVM heap sizes to at least 50% of system RAM
-      - cluster.routing.allocation.disk.threshold_enabled=false # Avoid ES going read-only because low disk space availability
-      - "DISABLE_SECURITY_PLUGIN=true"     # Disable security plugin
-      - plugins.ml_commons.allow_registering_model_via_url=true
-      - discovery.seed_hosts=search        # Nodes to look for when discovering the cluster
+      - "OPENSEARCH_JAVA_OPTS=-Xms1g -Xmx1g" # Set min and max JVM heap sizes to at least 50% of system RAM
+      - "DISABLE_SECURITY_PLUGIN=true"       # Disable security plugin
+      - cluster.name=os-docker-cluster       # Search cluster name
+      - cluster.routing.allocation.disk.threshold_enabled=false     # Avoid ES going read-only because low disk space availability
       - cluster.initial_cluster_manager_nodes=opensearch-node-data # Nodes eligible to serve as cluster manager
+      - node.name=opensearch-node-data       # Name the node that will run in this container
+      - bootstrap.memory_lock=true           # Disable JVM heap memory swapping
+      - discovery.seed_hosts=search          # Nodes to look for when discovering the cluster
+      - plugins.ml_commons.allow_registering_model_via_url=true
+      - plugins.ml_commons.native_memory_threshold=100 # Prevent memory issue after multiple deploy (https://github.com/opensearch-project/ml-commons/issues/2308)
+      - plugins.ml_commons.jvm_heap_memory_threshold=100 # Prevent memory issue after multiple deploy (https://github.com/opensearch-project/ml-commons/issues/2308)
     volumes:
       - os2_data:/usr/share/opensearch/data:rw
     ulimits:
@@ -160,16 +162,18 @@ services:
       target: gally_opensearch2
     # restart: unless-stopped
     environment:
-      - cluster.name=os-docker-cluster     # Search cluster name
-      - node.name=opensearch-node-ml # Name the node that will run in this container
-      - bootstrap.memory_lock=true         # Disable JVM heap memory swapping
-      - "OPENSEARCH_JAVA_OPTS=-Xms512m -Xmx512m" # Set min and max JVM heap sizes to at least 50% of system RAM
+      - "OPENSEARCH_JAVA_OPTS=-Xms1g -Xmx1g" # Set min and max JVM heap sizes to at least 50% of system RAM
+      - "DISABLE_SECURITY_PLUGIN=true"       # Disable security plugin
+      - cluster.name=os-docker-cluster       # Search cluster name
       - cluster.routing.allocation.disk.threshold_enabled=false # Avoid ES going read-only because low disk space availability
-      - "DISABLE_SECURITY_PLUGIN=true"     # Disable security plugin
-      - node.roles=ml
-      - plugins.ml_commons.allow_registering_model_via_url=true
-      - discovery.seed_hosts=search # Nodes to look for when discovering the cluster
       - cluster.initial_cluster_manager_nodes=opensearch-node-data # Nodes eligible to serve as cluster manager
+      - node.name=opensearch-node-ml         # Name the node that will run in this container
+      - node.roles=ml                        # Define this node as an ml node
+      - bootstrap.memory_lock=true           # Disable JVM heap memory swapping
+      - discovery.seed_hosts=search          # Nodes to look for when discovering the cluster
+      - plugins.ml_commons.allow_registering_model_via_url=true
+      - plugins.ml_commons.native_memory_threshold=100 # Prevent memory issue after multiple deploy (https://github.com/opensearch-project/ml-commons/issues/2308)
+      - plugins.ml_commons.jvm_heap_memory_threshold=100 # Prevent memory issue after multiple deploy (https://github.com/opensearch-project/ml-commons/issues/2308)
     ulimits:
       memlock:
         soft: -1