dummy public msg required by parent CL

perfkitbenchmarker/benchmark_spec.py

@@ -54,7 +54,6 @@
 from perfkitbenchmarker import providers
 from perfkitbenchmarker import relational_db
 from perfkitbenchmarker import resource as resource_type
-from perfkitbenchmarker import resources  # pylint:disable=unused-import  # Load the __init__.py
 from perfkitbenchmarker import smb_service
 from perfkitbenchmarker import stages
 from perfkitbenchmarker import static_virtual_machine as static_vm

perfkitbenchmarker/configs/benchmark_config_spec.py

@@ -40,8 +40,6 @@
 from perfkitbenchmarker.configs import vm_group_decoders
 from perfkitbenchmarker.resources import example_resource_spec
 from perfkitbenchmarker.resources import jobs_setter
-# Included to import & load Kubernetes' __init__.py somewhere.
-from perfkitbenchmarker.resources import kubernetes  # pylint:disable=unused-import
 from perfkitbenchmarker.resources import managed_ai_model_spec
 from perfkitbenchmarker.resources.pinecone import pinecone_resource_spec
 from perfkitbenchmarker.resources.vertex_vector_search import vvs_resource_spec

perfkitbenchmarker/configs/default_benchmark_config.yaml

@@ -390,24 +390,6 @@ postgres_sysbench_tpcc:
     sysbench_scale: 100
     sysbench_use_fk: False
 
-mongodb:
-  name: mongodb_ycsb
-  flags:
-    mongodb_readahead_kb: 8
-    iostat: True
-    sar: True
-    sar_interval: 1
-    ycsb_fail_on_incomplete_loading: True
-    ycsb_measurement_type: hdrhistogram
-    ycsb_status: True
-    ycsb_status_interval_sec: 1
-    ycsb_operation_count: 1000000000
-    ycsb_record_command_line: False
-    ycsb_run_parameters: dataintegrity=true,readallfields=true,writeallfields=true
-    timeout_minutes: 360
-    mongodb_primary_only: True
-    mongodb_pss: False
-
 fio_latency:
   name: fio
   flags:
@@ -461,3 +443,74 @@ hadoop_dfsio:
           disk_size: 6830
           disk_type: hyperdisk-throughput
           provisioned_throughput: 600
+
+diskspd_read_only:
+  name: diskspd
+  flags: &diskspd_read_only_flags
+    diskspd_prefill_duration: 2700
+    diskspd_write_read_ratio: 0
+
+diskspd_write_only:
+  name: diskspd
+  flags: &diskspd_write_only_flags
+    diskspd_write_read_ratio: 100
+
+diskspd_single_thread_latency:
+  name: diskspd
+  flags: &diskspd_single_thread_latency_flags
+    diskspd_thread_number_per_file: 1
+    diskspd_outstanding_io: 1
+    diskspd_block_size: 4k
+    diskspd_access_pattern: r
+
+diskspd_max_iops:
+  name: diskspd
+  flags: &diskspd_max_iops_flags
+    diskspd_thread_number_per_file: 1,4,8,16
+    diskspd_outstanding_io: 128
+    diskspd_block_size: 4k
+    diskspd_access_pattern: r
+
+diskspd_max_throughput:
+  name: diskspd
+  flags: &diskspd_max_throughput_flags
+    diskspd_thread_number_per_file: 8,16
+    diskspd_outstanding_io: 64,128
+    diskspd_block_size:  1M
+    diskspd_access_pattern: s
+
+diskspd_single_thread_latency_read_only:
+  name: diskspd
+  flags:
+    <<: *diskspd_read_only_flags
+    <<: *diskspd_single_thread_latency_flags
+
+diskspd_single_thread_latency_write_only:
+  name: diskspd
+  flags:
+    <<: *diskspd_write_only_flags
+    <<: *diskspd_single_thread_latency_flags
+
+diskspd_max_iops_read_only:
+  name: diskspd
+  flags:
+    <<: *diskspd_read_only_flags
+    <<: *diskspd_max_iops_flags
+
+diskspd_max_iops_write_only:
+  name: diskspd
+  flags:
+    <<: *diskspd_write_only_flags
+    <<: *diskspd_max_iops_flags
+
+diskspd_max_throughput_read_only:
+  name: diskspd
+  flags:
+    <<: *diskspd_read_only_flags
+    <<: *diskspd_max_throughput_flags
+
+diskspd_max_throughput_write_only:
+  name: diskspd
+  flags:
+    <<: *diskspd_write_only_flags
+    <<: *diskspd_max_throughput_flags

perfkitbenchmarker/data/docker/flink/cloudbuild.yaml.j2

@@ -1,6 +1,6 @@
 steps:
-- name: gcr.io/cloud-builders/gsutil
-  args: ['cp', {{ dpb_job_jarfile }}, 'job.jar']
+- name: gcr.io/cloud-builders/gcloud
+  args: ['storage', 'cp', {{ dpb_job_jarfile }}, 'job.jar']
 - name: 'gcr.io/cloud-builders/docker'
   args: ['build',
          '--build-arg',

perfkitbenchmarker/data/spark/spark-defaults.conf.j2

@@ -17,6 +17,10 @@ spark.master=spark://{{ leader_ip }}:7077
 # Hadoop configs prefixed with spark.hadoop
 spark.hadoop.fs.defaultFS=hdfs://{{ leader_ip }}/
 
+{% if shuffle_partitions is not none %}
+spark.sql.shuffle.partitions={{ shuffle_partitions }}
+{% endif %}
+
 {% for key, value in spark_conf.items() %}
 {{ key }}={{ value }}
 {% endfor %}

perfkitbenchmarker/edw_service.py

@@ -180,6 +180,9 @@
 
 FLAGS = flags.FLAGS
 
+EDW_PYTHON_DRIVER_LIB_FILE = 'edw_python_driver_lib.py'
+EDW_PYTHON_DRIVER_LIB_DIR = 'edw/common/clients/python'
+
 TYPE_2_PROVIDER = dict([
     ('athena', 'aws'),
     ('redshift', 'aws'),

perfkitbenchmarker/linux_benchmarks/dpb_sparksql_benchmark.py

@@ -45,6 +45,7 @@
 """
 
 from collections.abc import MutableMapping
+import functools
 import json
 import logging
 import os
@@ -53,11 +54,13 @@
 from typing import Any, List
 
 from absl import flags
+from perfkitbenchmarker import background_tasks
 from perfkitbenchmarker import configs
 from perfkitbenchmarker import dpb_constants
 from perfkitbenchmarker import dpb_service
 from perfkitbenchmarker import dpb_sparksql_benchmark_helper
 from perfkitbenchmarker import errors
+from perfkitbenchmarker import linux_virtual_machine
 from perfkitbenchmarker import object_storage_service
 from perfkitbenchmarker import sample
 from perfkitbenchmarker import temp_dir
@@ -123,6 +126,10 @@
     ' DPB services.',
 )
 
+_READAHEAD_KB = flags.DEFINE_integer(
+    'sparksql_readahead_kb', None, 'Configure block device readahead settings.'
+)
+
 FLAGS = flags.FLAGS
 
 LOG_RESULTS_PATTERN = (
@@ -214,6 +221,14 @@ def CheckPrerequisites(benchmark_config):
     )
 
 
+def _PrepareNode(vm: linux_virtual_machine.BaseLinuxVirtualMachine) -> None:
+  if _READAHEAD_KB.value is not None:
+    vm.SetReadAhead(
+        _READAHEAD_KB.value * 2,
+        [d.GetDevicePath() for d in vm.scratch_disks],
+    )
+
+
 def Prepare(benchmark_spec):
   """Installs and sets up dataset on the Spark clusters.
 
@@ -223,6 +238,12 @@ def Prepare(benchmark_spec):
   Args:
     benchmark_spec: The benchmark specification
   """
+  # Only unmanaged dpb services are VM-aware
+  if benchmark_spec.dpb_service.CLOUD == 'Unmanaged':
+    nodes = benchmark_spec.dpb_service.vms['worker_group']
+    partials = [functools.partial(_PrepareNode, node) for node in nodes]
+    background_tasks.RunThreaded((lambda f: f()), partials)
+
   cluster = benchmark_spec.dpb_service
   storage_service = cluster.storage_service
 

perfkitbenchmarker/linux_benchmarks/mongodb_ycsb_benchmark.py

@@ -40,11 +40,11 @@
 from perfkitbenchmarker.linux_packages import ycsb
 
 flags.DEFINE_integer(
-    'mongodb_readahead_kb', None, 'Configure block device readahead settings.'
+    'mongodb_readahead_kb', 8, 'Configure block device readahead settings.'
 )
 flags.DEFINE_bool(
     'mongodb_primary_only',
-    False,
+    True,
     'Run with a simple primary-only setup. Mutually exclusive with'
     ' --mongodb_pss. If both are False, the default PSA setup will be used.',
 )
@@ -78,7 +78,19 @@
   description: Run YCSB against MongoDB.
   vm_groups:
     primary:
-      vm_spec: *default_dual_core
+      vm_spec:
+        GCP:
+          machine_type: n4-standard-2
+          zone: us-central1-b
+          boot_disk_size: 100
+        Azure:
+          machine_type: Standard_D2s_v6
+          zone: eastus-1
+          boot_disk_size: 100
+        AWS:
+          machine_type: m7i.large
+          zone: us-east-1a
+          boot_disk_size: 100
       disk_spec:
         GCP:
           disk_size: 500
@@ -94,7 +106,19 @@
           mount_point: /scratch
       vm_count: 1
     secondary:
-      vm_spec: *default_dual_core
+      vm_spec:
+        GCP:
+          machine_type: n4-standard-2
+          zone: us-central1-b
+          boot_disk_size: 100
+        Azure:
+          machine_type: Standard_D2s_v6
+          zone: eastus-1
+          boot_disk_size: 100
+        AWS:
+          machine_type: m7i.large
+          zone: us-east-1a
+          boot_disk_size: 100
       disk_spec:
         GCP:
           disk_size: 500
@@ -110,7 +134,19 @@
           mount_point: /scratch
       vm_count: 1
     secondary_2:
-      vm_spec: *default_dual_core
+      vm_spec:
+        GCP:
+          machine_type: n4-standard-2
+          zone: us-central1-b
+          boot_disk_size: 100
+        Azure:
+          machine_type: Standard_D2s_v6
+          zone: eastus-1
+          boot_disk_size: 100
+        AWS:
+          machine_type: m7i.large
+          zone: us-east-1a
+          boot_disk_size: 100
       disk_spec:
         GCP:
           disk_size: 500
@@ -138,6 +174,27 @@
     fstab_options: noatime
     enable_transparent_hugepages: false
     create_and_boot_post_task_delay: 5
+    ycsb_fail_on_incomplete_loading: True
+    ycsb_measurement_type: hdrhistogram
+    ycsb_status: True
+    ycsb_status_interval_sec: 1
+    ycsb_operation_count: 1000000000
+    ycsb_record_command_line: False
+    ycsb_run_parameters: dataintegrity=true,readallfields=true,writeallfields=true
+    ycsb_client_vms: 1
+    ycsb_preload_threads: 512
+    ycsb_threads_per_client: 2048
+    ycsb_workload_files: workloadac,workloada,workloadx
+    ycsb_field_count: 10
+    ycsb_field_length: 100
+    ycsb_record_count: 20000000
+    ycsb_sleep_after_load_in_sec: 300
+    ycsb_timelimit: 300
+    ycsb_sleep_between_thread_runs_sec: 120
+    os_type: ubuntu2204
+    timeout_minutes: 360
+    sar: True
+    sar_interval: 1
 """
 
 _LinuxVM = linux_virtual_machine.BaseLinuxVirtualMachine

perfkitbenchmarker/linux_benchmarks/unmanaged_postgresql_sysbench_benchmark.py

@@ -20,6 +20,7 @@
 
 import copy
 import logging
+import re
 import time
 
 from absl import flags
@@ -105,10 +106,27 @@
 _OLTP_WRITE_ONLY = 'oltp_write_only'
 _OLTP = [_OLTP_READ_WRITE, _OLTP_READ_ONLY, _OLTP_WRITE_ONLY]
 
-_SHARED_BUFFER_SIZE = flags.DEFINE_integer(
+_SHARED_BUFFER_SIZE = flags.DEFINE_string(
     'postgresql_shared_buffer_size',
-    10,
-    'Size of the shared buffer in the postgresql cluster (in Gb).',
+    '10G',
+    'Size of the shared buffer in the postgresql cluster.'
+    'Format: <size>[<unit>], where <unit> is one of (B, K, M, G). '
+    'Example: 16G, 512M. If no unit is specified, G is assumed by default.',
+)
+
+
+def _ValidateSharedBufferSizeFlagValue(value: str) -> bool:
+  """Validates the shared buffer size flag's format."""
+  # Checks for one or more digits, optionally followed by B, K, M, or G.
+  return bool(re.fullmatch(r'^\d+[BKMG]?$', value))
+
+flags.register_validator(
+    _SHARED_BUFFER_SIZE,
+    _ValidateSharedBufferSizeFlagValue,
+    message=(
+        '--postgresql_shared_buffer_size must be in the format <size>[<unit>] '
+        'where <unit> is one of (B, K, M, G). Example: 16G, 512M, 1024K, 2048B.'
+    )
 )
 _MEASURE_MAX_QPS = flags.DEFINE_bool(
     'postgresql_measure_max_qps',
@@ -123,6 +141,16 @@
 )
 
 
+def GetBufferSize() -> str:
+  """Returns the buffer key for the given buffer size."""
+  buffer_size = _SHARED_BUFFER_SIZE.value
+  if buffer_size.endswith(
+      ('B', 'K', 'M', 'G')
+  ):
+    return buffer_size
+  return f'{buffer_size}G'
+
+
 def GetConfig(user_config):
   """Get the benchmark config, applying user overrides.
 
@@ -185,7 +213,7 @@ def Prepare(benchmark_spec: bm_spec.BenchmarkSpec):
   postgresql.ConfigureAndRestart(
       primary_server,
       FLAGS.run_uri,
-      _SHARED_BUFFER_SIZE.value,
+      GetBufferSize(),
       _CONF_TEMPLATE_PATH.value,
   )
   for index, replica in enumerate(replica_servers):
@@ -194,7 +222,7 @@ def Prepare(benchmark_spec: bm_spec.BenchmarkSpec):
         replica,
         index,
         FLAGS.run_uri,
-        _SHARED_BUFFER_SIZE.value,
+        GetBufferSize(),
         _CONF_TEMPLATE_PATH.value,
     )
   clients = benchmark_spec.vm_groups['client']