[RLlib] Load state from load_state_path for rlmodule spec. (ray-proje…

…ct#35180)
krfricke · May 30, 2023 · 83179ab · 83179ab
1 parent 33adc0e
commit 83179ab
Show file tree

Hide file tree

Showing 15 changed files with 1,237 additions and 45 deletions.
diff --git a/release/release_tests.yaml b/release/release_tests.yaml
@@ -3458,6 +3458,34 @@
  cluster_env: app_config.yaml
  cluster_compute: multi_node_checkpointing_compute_config_gce.yaml
 
+- name: rllib_learner_e2e_module_loading
+ group: RLlib tests
+ working_dir: rllib_tests
+
+ frequency: nightly
+ team: rllib
+
+ cluster:
+ cluster_env: app_config.yaml
+ cluster_compute: multi_node_checkpointing_compute_config.yaml
+
+ run:
+ timeout: 3600
+ script: pytest checkpointing_tests/test_e2e_rl_module_restore.py
+
+ wait_for_nodes:
+ num_nodes: 3
+
+ alert: default
+
+ variations:
+ - __suffix__: aws
+ - __suffix__: gce
+ env: gce
+ frequency: manual
+ cluster:
+ cluster_env: app_config.yaml
+ cluster_compute: multi_node_checkpointing_compute_config_gce.yaml
 
 - name: rllib_multi_node_e2e_training_smoke_test
  group: RLlib tests
@@ -3488,8 +3516,6 @@
  cluster_env: app_config.yaml
  cluster_compute: multi_node_checkpointing_compute_config_gce.yaml
 
-
-
 - name: rllib_learning_tests_a2c_tf
  group: RLlib tests
  working_dir: rllib_tests