[ci][train] Remove unnecessary xgboost_ray/lightgbm_ray reinstall…

…s for release tests (#43176)

Pin `xgboost_ray` and `lightgbm_ray` versions to the current master commit in order to fix release tests that were still using the latest release.

---------

Signed-off-by: Justin Yu <[email protected]>

release/lightgbm_tests/workloads/tune_16x4.py

@@ -9,6 +9,7 @@
 as the Ray Tune results table. No trials should error. All trials should
 run in parallel.
 """
+
 from collections import Counter
 import json
 import os
@@ -58,6 +59,7 @@ def train_wrapper(config, ray_params):
  config=search_space,
  num_samples=16,
  resources_per_trial=ray_params.get_tune_resources(),
+ storage_path="/mnt/cluster_storage",
  )
  taken = time.time() - start
 

release/lightgbm_tests/workloads/tune_4x16.py

@@ -9,6 +9,7 @@
 as the Ray Tune results table. No trials should error. All trials should
 run in parallel.
 """
+
 from collections import Counter
 import json
 import os
@@ -58,6 +59,7 @@ def train_wrapper(config, ray_params):
  config=search_space,
  num_samples=4,
  resources_per_trial=ray_params.get_tune_resources(),
+ storage_path="/mnt/cluster_storage",
  )
  taken = time.time() - start
 

release/lightgbm_tests/workloads/tune_small.py

@@ -9,6 +9,7 @@
 as the Ray Tune results table. No trials should error. All trials should
 run in parallel.
 """
+
 from collections import Counter
 import json
 import os
@@ -58,6 +59,7 @@ def train_wrapper(config, ray_params):
  config=search_space,
  num_samples=4,
  resources_per_trial=ray_params.get_tune_resources(),
+ storage_path="/mnt/cluster_storage",
  )
  taken = time.time() - start
 

release/ray_release/byod/byod_xgboost_cuj_test.sh

@@ -4,4 +4,4 @@
 
 set -exo pipefail
 
-pip3 install -U --force-reinstall --no-deps xgboost_ray
+pip install -U "git+https://github.com/ray-project/xgboost_ray@5a840af05d487171883dadbfdd37b138b607bed8#egg=xgboost_ray"

release/ray_release/byod/byod_xgboost_test.sh

@@ -4,7 +4,7 @@
 
 set -exo pipefail
 
-pip3 install -U --force-reinstall --no-deps xgboost_ray lightgbm_ray
+pip3 install -U --force-reinstall --no-deps "git+https://github.com/ray-project/xgboost_ray@5a840af05d487171883dadbfdd37b138b607bed8#egg=xgboost_ray" "git+https://github.com/ray-project/lightgbm_ray@4c4d3413f86db769bddb6d08e2480a04bc75d712#egg=lightgbm_ray"
 sudo mkdir -p /data || true
 sudo chown ray:1000 /data || true
 rm -rf /data/classification.parquet || true

release/ray_release/byod/byod_xgboost_tune_test.sh

@@ -4,7 +4,7 @@
 
 set -exo pipefail
 
-pip3 install -U --force-reinstall --no-deps xgboost_ray
+pip3 install -U --force-reinstall --no-deps "git+https://github.com/ray-project/xgboost_ray@5a840af05d487171883dadbfdd37b138b607bed8#egg=xgboost_ray" "git+https://github.com/ray-project/lightgbm_ray@4c4d3413f86db769bddb6d08e2480a04bc75d712#egg=lightgbm_ray"
 sudo mkdir -p /data || true
 sudo chown ray:1000 /data || true
 rm -rf /data/train.parquet || true

release/release_tests.yaml

@@ -1647,35 +1647,6 @@
 
  alert: default
 
-- name: ml_user_xgboost_gpu_connect_latest
- group: ML user tests
- working_dir: ml_user_tests
-
- frequency: nightly-3x
- team: ml
-
- cluster:
- byod:
- type: gpu
- post_build_script: byod_xgboost_test.sh
- cluster_compute: xgboost/tpl_gpu_small_scaling_aws.yaml
-
- run:
- timeout: 1200
- script: python xgboost/train_gpu_connect.py
- wait_for_nodes:
- num_nodes: 5
-
- variations:
- - __suffix__: aws
- - __suffix__: gce
- env: gce
- frequency: manual
- cluster:
- cluster_compute: xgboost/tpl_gpu_small_scaling_gce.yaml
-
- alert: default
-
 - name: ml_user_xgboost_gpu_connect_master
  group: ML user tests
  working_dir: ml_user_tests
@@ -1686,7 +1657,7 @@
  cluster:
  byod:
  type: gpu
- post_build_script: byod_xgboost_master_test.sh
+ post_build_script: byod_xgboost_test.sh
  cluster_compute: xgboost/tpl_gpu_small_scaling_aws.yaml
 
  run:

release/xgboost_tests/workloads/tune_32x4.py

@@ -9,6 +9,7 @@
 as the Ray Tune results table. No trials should error. All trials should
 run in parallel.
 """
+
 from collections import Counter
 import json
 import os
@@ -58,6 +59,7 @@ def train_wrapper(config, ray_params):
  config=search_space,
  num_samples=32,
  resources_per_trial=ray_params.get_tune_resources(),
+ storage_path="/mnt/cluster_storage",
  )
  taken = time.time() - start
 

release/xgboost_tests/workloads/tune_4x32.py

@@ -9,6 +9,7 @@
 as the Ray Tune results table. No trials should error. All trials should
 run in parallel.
 """
+
 from collections import Counter
 import json
 import os
@@ -58,6 +59,7 @@ def train_wrapper(config, ray_params):
  config=search_space,
  num_samples=4,
  resources_per_trial=ray_params.get_tune_resources(),
+ storage_path="/mnt/cluster_storage",
  )
  taken = time.time() - start
 

release/xgboost_tests/workloads/tune_small.py

@@ -9,6 +9,7 @@
 as the Ray Tune results table. No trials should error. All trials should
 run in parallel.
 """
+
 from collections import Counter
 import json
 import os
@@ -58,6 +59,7 @@ def train_wrapper(config, ray_params):
  config=search_space,
  num_samples=4,
  resources_per_trial=ray_params.get_tune_resources(),
+ storage_path="/mnt/cluster_storage",
  )
  taken = time.time() - start