chore: Migrate gsutil usage to gcloud storage

perfkitbenchmarker/archive.py

@@ -28,7 +28,7 @@ def ArchiveRun(
     run_temp_directory,
     target_bucket,
     prefix='',
-    gsutil_path='gsutil',
+    gsutil_path='gcloud',
     aws_path=AWS_PATH,
 ):
   """Archive a run directory to GCS or S3.
@@ -52,7 +52,7 @@ def ArchiveRun(
   )
 
   prefix_len = 5
-  prefixes = {'s3://': [aws_path, 's3', 'cp'], 'gs://': [gsutil_path, 'cp']}
+  prefixes = {'s3://': [aws_path, 's3', 'cp'], 'gs://': [gsutil_path, 'storage', 'cp']}
 
   assert all(len(key) == prefix_len for key in prefixes), prefixes
 

perfkitbenchmarker/data/beam/flink-presubmit.sh

@@ -1,5 +1,5 @@
 #!/bin/bash
 echo 'Executing $0'
 echo 'Copying job jar file $1 into VM...'
-gsutil cp "$1" ./job.jar
+gcloud storage cp "$1" ./job.jar
 echo 'Finished copying job jar file.'

perfkitbenchmarker/linux_benchmarks/gcsfuse_benchmark.py

@@ -118,7 +118,7 @@ def _ReadThroughputTest(vm, mountpoint):
   """Read the files in the directory via tf.io.gfile or gcsfuse."""
   data_dir = FLAGS.gcsfuse_data_dir
   options = f'--mountpoint="{mountpoint}"'
-  cmd = f'gsutil ls "{data_dir}" | {_DLVM_PYTHON} {_REMOTE_SCRIPT} {options}'
+  cmd = f'gcloud storage ls "{data_dir}" | {_DLVM_PYTHON} {_REMOTE_SCRIPT} {options}'
   logging.info(cmd)
   stdout, stderr = vm.RemoteCommand(cmd)
   logging.info(stdout)

perfkitbenchmarker/linux_benchmarks/resnet_benchmark.py

@@ -188,7 +188,7 @@ def Prepare(benchmark_spec):
     vm.RemoteCommand(
         'sudo mkdir -p {data_path} && '
         'sudo chmod a+w {data_path} && '
-        'gsutil -m cp -r {data_dir}/* {data_path}'.format(
+        'gcloud storage cp --recursive {data_dir}/* {data_path}'.format(
             data_dir=benchmark_spec.data_dir, data_path=local_data_path
         )
     )

perfkitbenchmarker/linux_packages/tcmalloc.py

@@ -86,7 +86,7 @@ def AptInstall(vm):
     vm.Install('google_cloud_sdk')
     local_path = '/tmp/libtcmalloc.so'
     vm.RemoteCommand(
-        'gsutil cp {url} {path} '
+        'gcloud storage cp {url} {path} '
         '&& echo "export LD_PRELOAD={path}" | sudo tee -a {tmp}'.format(
             url=FLAGS.tcmalloc_experimental_url,
             path=local_path,

perfkitbenchmarker/log_collector.py

@@ -65,13 +65,13 @@ def CollectPKBLogs(run_uri: str, log_local_path: str) -> None:
     # Generate the log path to the cloud bucket based on the invocation date of
     # this function.
     gcs_log_path = GetLogCloudPath(PKB_LOG_BUCKET.value, f'{run_uri}-pkb.log')
+    # The gsutil top-level flag '-h' is not supported by 'gcloud storage cp' or 'gcloud storage mv'.
     vm_util.IssueCommand(
         [
-            'gsutil',
-            '-h',
-            'Content-Type:text/plain',
+            'gcloud',
+            'storage',
             _SAVE_LOG_TO_BUCKET_OPERATION.value,
-            '-Z',
+            '--gzip-local-all',
             log_local_path,
             gcs_log_path,
         ],
@@ -94,13 +94,13 @@ def CollectVMLogs(run_uri: str, source_path: str) -> None:
     source_filename = source_path.split('/')[-1]
     gcs_directory_path = GetLogCloudPath(VM_LOG_BUCKET.value, run_uri)
     gcs_path = f'{gcs_directory_path}/{source_filename}'
+    # The gsutil top-level flag '-h' is not supported by 'gcloud storage mv'.
     vm_util.IssueCommand(
         [
-            'gsutil',
-            '-h',
-            'Content-Type:text/plain',
+            'gcloud',
+            'storage',
             'mv',
-            '-Z',
+            '--gzip-local-all',
             source_path,
             gcs_path,
         ],

perfkitbenchmarker/object_storage_service.py

@@ -185,7 +185,7 @@ def MakeBucket(self, bucket, raise_on_failure=True, tag_bucket=True):
   def Copy(self, src_url, dst_url, recursive=False):
     """Copy files, objects and directories.
 
-    Note: Recursive copy behavior mimics gsutil cp -r where:
+    Note: Recursive copy behavior mimics gcloud storage cp --recursive where:
     Copy(/foo/bar, /baz, True) copies the directory bar into /baz/bar whereas
     aws s3 cp --recursive would copy the contents of bar into /baz.
 

perfkitbenchmarker/providers/README.md

@@ -427,6 +427,7 @@ To create a new `~/.boto` file, issue the following command and follow the
 instructions given by this command:
 
 ```bash
+# "gsutil config" has no direct "gcloud storage" equivalent. Its functionality is handled by "gcloud auth" and "gcloud config".
 $ gsutil config
 ```
 

perfkitbenchmarker/providers/gcp/gce_windows_virtual_machine.py

@@ -27,7 +27,7 @@
 from perfkitbenchmarker.providers.gcp import gcs
 from perfkitbenchmarker.providers.gcp import util
 
-_WINDOWS_SHUTDOWN_SCRIPT_PS1 = 'Write-Host | gsutil cp - {preempt_marker}'
+_WINDOWS_SHUTDOWN_SCRIPT_PS1 = 'Write-Host | gcloud storage cp - {preempt_marker}'
 
 _METADATA_PREEMPT_CMD_WIN = (
     'Invoke-RestMethod -Uri'

perfkitbenchmarker/scripts/gcsfuse_scripts/read.py

@@ -3,11 +3,11 @@
 Example usages:
 
 (1) Specify `--mountpoint` to read from gcsfuse.
-> gsutil ls gs://gcsfuse-benchmark/10M/ | python read.py --mountpoint=/gcs/
+> gcloud storage ls gs://gcsfuse-benchmark/10M/ | python read.py --mountpoint=/gcs/
 
 (2) Omit `--mountpoint` to read from GCS using tf.io.gfile; specify
 `--iterations` to run it multiple times.
-> gsutil ls gs://gcsfuse-benchmark/10M/ | python read.py --iterations=3
+> gcloud storage ls gs://gcsfuse-benchmark/10M/ | python read.py --iterations=3
 """
 
 import concurrent.futures