Add support for preconditioning scratch disks using fio before formatting.

perfkitbenchmarker/disk_strategies.py

@@ -329,6 +329,29 @@ def SetUpDiskOnLinux(self):
     self.vm.scratch_disks.append(smb_disk)
 
 
+def _PreconditionDisk(vm, scratch_disk):
+  """Preconditions a raw disk device using fio."""
+  # pylint: disable=g-import-not-at-top
+  from perfkitbenchmarker import linux_packages
+  logging.info(
+      'Preconditioning disk %s on %s', scratch_disk.GetDevicePath(), vm.name
+  )
+  # Ensure the install directory exists
+  vm.RemoteCommand(
+      f'sudo mkdir -p {linux_packages.INSTALL_DIR} && '
+      f'sudo chmod a+rwxt {linux_packages.INSTALL_DIR}'
+  )
+  vm.Install('fio')
+  device_path = scratch_disk.GetDevicePath()
+  fio_path = f'{linux_packages.INSTALL_DIR}/fio/fio'
+  cmd = (
+      f'sudo {fio_path} --filename={device_path} --ioengine=libaio '
+      f'--name=precondition --blocksize=1M --iodepth=64 '
+      f'--rw=write --direct=1 --fallocate=none --size=100%'
+  )
+  vm.RobustRemoteCommand(cmd)
+
+
 class PrepareScratchDiskStrategy:
   """Strategies to prepare scratch disks."""
 
@@ -375,6 +398,8 @@ def PrepareLinuxScratchDisk(
     if disk_spec.mount_point and not vm.IsMounted(
         disk_spec.mount_point, scratch_disk.GetDevicePath()
     ):
+      if FLAGS.precondition_disks:
+        _PreconditionDisk(vm, scratch_disk)
       vm.FormatDisk(scratch_disk.GetDevicePath(), disk_spec.disk_type)
       vm.MountDisk(
           scratch_disk.GetDevicePath(),

perfkitbenchmarker/linux_packages/postgresql.py

@@ -62,6 +62,18 @@
         'max_memory': '108.8G',
         'nr_hugepages': '23000',
     },
+    'SIZE_64G': {  # 3TB DB (200 tables, 60M tablesize) for read/write
+        'shared_buffers': '64GB',
+        'effective_cache_size': '94GB',
+        'max_memory': '120G',
+        'nr_hugepages': '35000',
+    },
+    'SIZE_32G': {  # 3TB DB (200 tables, 60M tablesize) for read-only
+        'shared_buffers': '32GB',
+        'effective_cache_size': '94GB',
+        'max_memory': '120G',
+        'nr_hugepages': '25000',
+    },
 }
 
 POSTGRES_CLUSTER_NAME = 'main'

perfkitbenchmarker/linux_virtual_machine.py

@@ -154,6 +154,11 @@
     [_DEFAULT_DISK_FS_TYPE, 'xfs'],
     'File system type used to format disk.',
 )
+flags.DEFINE_bool(
+    'precondition_disks',
+    False,
+    'Precondition scratch disks using fio before formatting.',
+)
 flags.DEFINE_integer(
     'disk_block_size',
     None,
@@ -582,6 +587,7 @@ def _SetupRobustCommand(self):
       if not self._has_remote_command_script:
         # Python is needed for RobustRemoteCommands
         self.Install('python')
+        self._CreateVmTmpDir()
 
         for f in (EXECUTE_COMMAND, WAIT_FOR_COMMAND):
           remote_path = os.path.join(vm_util.VM_TMP_DIR, os.path.basename(f))
@@ -1334,9 +1340,16 @@ def FormatDisk(self, device_path, disk_type=None):
       fmt_cmd = 'sudo mkfs.xfs -f -i size={} {}'.format(block_size, device_path)
     else:
       block_size = FLAGS.disk_block_size or 4096
+      mke2fs_options = 'lazy_itable_init=0'
+      if FLAGS.precondition_disks:
+        mke2fs_options += ',nodiscard'
+      else:
+        mke2fs_options += ',discard'
       fmt_cmd = (
-          'sudo mke2fs -F -E lazy_itable_init=0,discard -O '
-          '^has_journal -t ext4 -b {} {}'.format(block_size, device_path)
+          'sudo mke2fs -F -E {} -O '
+          '^has_journal -t ext4 -b {} {}'.format(
+              mke2fs_options, block_size, device_path
+          )
       )
     self.os_metadata['disk_filesystem_type'] = FLAGS.disk_fs_type
     self.os_metadata['disk_filesystem_blocksize'] = block_size